At first glance, the way collectl calculates its numbers is pretty straight forward. It
looks at successive intervals of counters, calculates their differences and normalizes
the result by dividing by the interval, the result of which is the counter's rate/second.
If -on is specified collectl does not divide by the interval and simply reports the difference.
However, one occasionally may see numbers that don't make sense, such as a 1Gb network
reporting rates almost double what it is capable of or
other anomolous numbers.
The Interval Time Stamps
By design, collectl takes one time stamp at the start of each monitoring interval and
associates that time with all the samples taken during that interval.
This has been done for one major reason - there needs to be a single time
associated with all data points, especially if you want to plot the data.
The overhead in collecting the data is fairly constant and therefore the interval for
that sample is fairly consistent and so the rates reported are also consistent.
However, there can be a problem that is important to understand and has been seen in
the past. A device had the wrong firmware
level and under some conditions caused a long delay in the middle of the collection
interval. Some samples were collected close the the starting time of that
interval while all that followed the delay were actually collected at a time much later
than was being reported.
Consider the following in which we're looking at raw data collected for 2 subsystems,
call them XXX and YYY. Let's also assume that the counters we're monitoring are
increasing at a steady rate of 100 units/sec. In this example, during the 10:00:01 interval
there was a 10 second hang in collecting the YYY sample. The XXX sample was correctly
recorded, but by the time the YYY sample was collected, 1000 units were recorded. As we
move to the next interval which was delayed by 10 seconds, the sample for XXX has
accumulated 1000 units and the sample for YYY is 100.
The problem here is when reporting the 2 rates at 10:00:01, we'll see a rate of 1000
units/sec for YYY because based on the timestamp that interval only appears to be 1
second long. Conversely, the rate reported for that same subsystem at 10:00:11 will be
10 units/sec because this interval is reported as 10 seconds long. Also note that for
this interval the counter for XXX has been incremented correctly and the resultant rates
are reported correctly.
This is because the sampling occured before the delays. If
one were to move the timestamp to the end of the interval, it would fix the problem with
YYY, but then move it to XXX.
It IS important to understand that this is only a problem if the delay is during the data
collection itself. If there is a system delay that causes all data collection to be
delayed but once started runs as expected, and this has been seen to be the typical
case, the intervals may be longer but the counters will
have increased proportionaly and the results consistent.
The only real answer to this problem would be to timestamp individual samples, however
it is also felt that this problem is rare enough as to not be of serious concern and
changing the methodology of timestamping would cause more problems than it solves.
The Counter Update Rate
This is a problem that is very real and worth understanding even it if you never
personally see it. If the rate at which a counter is updated is too coarse,
especially if it is close to the monitoring interval, the reported numbers will
be off. For most of the data collectl reports on, this is not a problem because these
counters ARE updated frequently. However, it turns out that some network drivers only
updated about once a second and in early versions of the 2.6 kernel (and may the one
you're currently running on), you may see some very strange anomolies in the output if
you look at 1-second data samples.
See this page for more details.
As they say, garbage in, garbage out and so if the number you're seeing look wrong, it's
worth trying to understand why and you shouldn't necessary take them as face
value.
updated Feb 21, 2011
collectl-3.6.9/docs/RunningAsAService.html 0000664 0001750 0001750 00000011611 12230241727 016644 0 ustar mjs mjs
collectl - Running as a Service
Running As A Service
Assuming collectl has been installed from the rpm kit, it has been
configured to be run as a service, but disabled from automatically
starting at boot. To enable it, simply
chkconfig collectl on, noting that by default collectl is
configured to collect most data.
To see what the specific subsystems are, execute collectl -V
and look at the daemon default values for -s. You should then look at
the DaemonCommands string /etc/collectl.conf to see if any changes to
-s have been explictly set. At the time of this writing, collectl has been
further configured to add slab and process data to the base defaults.
Further inspection of this command string will show the daemon has also been
configured to write all its data to a set of compressed text files in
/var/log/collectl, which was created when the kit was installed. To verify
collectl will properly run as a service, you can execute the command
/etc/init.d/collectl start (or as a shortcut on a redhat system use
the command service collectl start) and examine the log file in /var/log/collectl
for the startup (and hopefully no termination) messages as well as the appearance
of either a raw or raw.gz data file in that same directory.
Note that since the output is buffered, the data file will probably have a length of
0 until the buffer fills or the flush interval passes, which is currently set to 60 seconds,
which ever comes first. Or the command /etc/init.d/collectl flush is executed.
In order to write its output as a compressed file, the perl Compress module
must be present as it is with newer perl distributions. If not present
you should install it, otherwise you will get messages warning you that compression is not
installed.
To change any behaviors of the daemon such as the flush interval, output file
location, etc., simply change the DaemonCommands line in
/etc/collectl.conf, which specifies the actual command string collectl is passed at
startup. Use care in setting this string as incorrect settings may cause collectl
to abnormally exit and if it does, you should examine the log file for
messages.
caution about pipes in DaemonCommands
Since some of the filters can include pipes, one might choose the use the perl form of
"abc|def|xyz" when using them interactively, having to use quotation marks to
prevent the shell from acting on them. However if you include the quotes in the
DaemonCommands line, the filters will not work correctly as collectl will see
the quotes as part of the filter itself.
One-time modification of runtime parameters
If you want to change the way collectl runs as a daemon for a specific instance, you can
pass the normal collectl switches to the start script as its second parameter (more on the
first parameter later). For example, to start collectl with a monitoring interval of
15 seconds, just start it as follows:
/etc/init.d/collectl start '-i 15'
The next time it is started (or restarted) it will use its default values.
Running multiple instances of a collectl daemon
By default, collectl only supports running a single instance of a daemon and it you try to
start a second you will get an error message. However, there may be times you really want to
run a second instance, most typically if you want to collect a subset of data at a different
monitoring interval, and to do this one uses an alternative syntax which prefaces the
parameters with a string such as test as in the following example:
/etc/init.d/collectl start test -i15
Also note in this example quotes weren't needed because there were no spaces in the second
argument. In this case a process named collectl-test will be created and use the
argument -i15. You must be careful when using this format because if you leave off the
second argument you'd actually start the main process with the invalid switch of test.
To perform other operations on this second instance, such as stop or flush, simply
add the test qualifier to the command. If you want to restart one of these instances be
sure to include the appropriate arguments because you must use the 2 argument form of the command.
This syntax was also chosen to assure the user does use additional switches because without them
you'd essentially be running an identical copy of the default configuration.
updated Sep 5, 2013
collectl-3.6.9/docs/Ganglia3.jpg 0000644 0001750 0001750 00000025363 12230241726 014565 0 ustar mjs mjs JFIF ` ` C
$.' ",# (7),01444 '9=82<.342 C
2! !22222222222222222222222222222222222222222222222222 @"
} !1AQa"q2#BR$3br
%&'()*456789:CDEFGHIJSTUVWXYZcdefghijstuvwxyz
w !1AQaq"2B #3Rbr
$4%&'()*56789:CDEFGHIJSTUVWXYZcdefghijstuvwxyz ? (
(
(
(
(
(
(,mu TXd8<@E QE QE QE QEeh& s!ӯ _5@%pNG:P uZI%"HG<@X/gc*WfZ
Hp2q\ g ]鹱Eq7?< gKS$WppAmqǽvqK ȲE"GC#jGaQE QE QPYues
ͻd py'( ( ( ( ( ( ( ( (m{w `tW^H>א [5 $u#Jf* Ak+>Icj rGmbm 3ȗggj] Hzi~]Ǐ| ii cĖOwr$K
IE; <;,+ogN[.|t?+ {}8Z._} F*
gm]B燑_CQ2!C
#dԼuZ-R Il,^_u9=V-
>֍qɨjRԮ-K:QN3RA@'zn Ymfv/"az 0iFN/-Knn X?|'o^la(y,sm `qqӵ$ y/~4 Cx72jv^rH ;B$᩵J@V:uZxygL{gGϧ Q,-n{ _ׯSnn|uhQm~lّ ^^Yhw:> ZMyZ9x4IB!K4* ؎B>YGzFA4]dJH{ҍ}e(0( ( ( ( ( ( ( ( ( ( ( ߋuxB
WLxc&u!R=^5W|}K úb#OVP>1
|ZV.Ė14R0k>j ֵχzϧLY|Dcp?>4QE QE QE QE QE QE QE QE QE QE QE QE QE QE QE QE QE QE QE n<7t6*x?PY i}+تZ xh 6}o -bYbqXd:^ISWM[z#Pfc? nqn
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
]_-U {`B~+ K((((((+/YKn&$G33?:i6ZRdxO// G, [ % j1sǹQ\, [ % h ? 伿M }sǹQ\, [ % h ? 伿M }sǹQ\, [ % h ? 伿M }sǹQ\, [ % h ? 伿M }sǹQ\, [ % h ? 伿M }sǹQ\, [ % h ? 伿M }sǹQ\, [ % h ? 伿M }sǹQ\, [ % h ? 伿M }sǹQ\, [ % h ? 伿M }sǹQ\, [ % h ? 伿M }sǹQ\, [ % h ? 伿M }sǹ'Pz7?p Nv^^W]}.ojo/*<}_
>*>
4v.lzg=sǹ4W+ # y?dxO// Gf uTW+ # y?dxO// Gf uTW+ # y?dxO// Gf uTW+ # y?dxO// Gf uTW+ # yU6`V' 綏g>9ho{ k]q. ֮ '=̻ETQE f
&\E`I$ ?ǦGQZUu{4^(0??(p{Vׁ|ik-P! wmy;PSEPEPEPHHU, I=kWOmwq ! ?,RG_u4>ks^u
Hcfa ?qhVW a !O }~}+J< uo" K?/2>P]Er$VQx~x^|)) }f!$ ?/B+JWӠ,&YP j EPExOBTlb 4z( < FΟ~~|HTn+}xzKv~z|;;Ay=Oa<N洅DG Ak( ( ( ( ( ( (ls 3Y#ER\葏GZ*xm K=
8ݗq}G?Uo/%edy| ^((( ȟ l bA\ ?O 1jkԙ,j(38x[ Zg\gp 赫Do.EW^^j3MoXR)ʨPx WF *ˋ iSch) yڇu?
jɭ oP[[X&RF![(%K?u3,mw#E^ OS}bK+Bp{?kwAa<9IbuxC#`y J}|گEEd4w zcGmj ):YuJF=fFzݽwr6y$c¨kɴ[K- RZsq;߱?\XmTKw$V #~{Wag}VzK#નyuZz.aNjzJ+|C:5 v AO ɭ O?j+|C:5G5Σ 8u<3t݅* &zKFI=>iXm3 ^cVv=+7\w)B.9 SWJ5εXѦKe^Sj ) d#`yǫQ^Sj ) d`yǫQ^Sj ) d`yǫWl ZÎ4`}?u_( Ǘ Ryڇu?
jִ95U%"'>ԞU]-,5 qWu?
jv AO ;P Χ MGMn? 'z>v AO _ivW61#&u:88ӾM\ ?O 1k ' ?O^|z5QPPW <- o-k38x[ Z "gho j 7^^S7o BS)3<> tk˘BU}M}i+RmQ]DŽ-.~G9$ zb ʐΊ,mir3䌌 Yk _Q^8wEt浜mar7 Zsk۟
[̏_OzO[NpvAb$('GAӿp+džn/!O4\O寙Fc?R25Vv{}ӂ&:7O]M}FԴI]ҬO뗋ޅXA q kwZ)ѥ]$ٕԏ=k"T]bYŷ _Ejƣhf_`]y-H$vJg' xEMne %OdIY/[ Cu s6U5Cu s6UcGGh?.+B t @_/GdB/K]q>< m F cSќQ_Z~rU J;KT+'үvnl|<$bXF\m
'8V.g/-3 jC "} @ZonW}lh!hdaQ].\j0ԢbK#mHǰGCӒ)|{%m-bֈWnj
Jsje/ [<ޠ# =jֹ}> #2"NYbp]꿗JRrQNPmʮrVNJh?Zӟ<YS & VJ AZsoS F-tD g
> =((+w <- o-j3˴QEAA^S7o BV)Bڷ W |VׄO٘F,W8< Z|SKo!Gi젗Ż7?
9+]_ö?kŚW[QZ7W=EsQq qĪѮr烜|pqw9snrNs;!l<-zkJ_/C:i^Sv[Kz^<[iwIf5+ׂ<0 zU (-Rg7#' {p=MWCn.;\m1Iwgsd6A|':}8F-ts9rk9ѭIJj+]o.v-kYhwcmWF ;i nr`=G(ippK䃃 "m Xἵb]vvw6Cd ۮ4 <{w-$I< m F cSќQ_Z~rljR;|`{4,8'ڮsTT)jf GXN4YtPڐX%aNqqVk$V6e9FJ۷6\^vM 0)Nܜg7UK{'L6%Iwn4s-u{ jh ȒFe]x ( H9TujJº & VJ AZ3oS F-tD g
> =((+w <- o-j3˴QEAA^[ygC_*dfR שQ]|DK'./ T=M)uj
y:?jj~#xKH<= -ns|f@
=ǭu]ƛeQV>滏x:ک͉.; tt kSsy> T-TWݞS uj
Vջ# S ::6jz(֭Ϝ/Ozk]z< P xRNe$ܲW]Z}i-I,)I#qz^Co- }ᥟ:C6;"ԧuo_'׳RS VdD6Rm[WX%u !})k5k#8XáiH%j (1VA\omnҦXˑeE]*C:ԕjnz S uj
W[< S uj
Q[ S uj
Q[ S uj
Q[ S uj
Q[ S uj
Q[ S 5Ŷv.mwT
85Xb1+ǖHpr}B ȟ l bA\ ?O 1kƽN,j(38x[ Zg\gp 赫Do.EQ@ <]iK%e-2 \W?7v?CuGăH`z>5땁 CO$ 4;ºfnUp֘Ӱqɠ
(
(
(
(
(
(
(
(
(
q "} ы]s8 > ūRgѨo{ k]q. ֮ '=̻ETQE QE QE QE QE QE QE QE QE QE QE QE QE QE QE QE D g
q "} ыWO^gQEe> |@!wyeۜg+Riz3 o? ĿU ' @ i>\r' @ UQGv r' @ UQGv r' @ UQGv r' @ UQGv r' @ UQGv r' @ UQGv r' @ UQGv r' @ UQGv r' @ UQGv r' @ UQGv r' @ UQGv r' @ UQGv r' @ UQGv r' @ UQGv r' @ UQGv r' @ UQGv r' @ |89 ɉꨣOH
( collectl-3.6.9/docs/SlowProc.html 0000664 0001750 0001750 00000015600 12230241727 015070 0 ustar mjs mjs
collectl - Slow Proc Access?
Is Your Kernel Reading /proc Too Slowly?
Introduction
The 2.6.32 release of the kernel introduced a regression that causes the /proc to be read significantly
more slowly and with significantly more overhead on systems with high core counts. In fact this
overhead has been measured to be over a factor of 50 reading /proc/stat on a system with 8 sockets and
48 cores.
The good news is newer RedHat and SUSE distros have been updated to mitigate this problem, specifically
RHEL 6.2 and SLES SP1. As for other distros, I just don't have the access to verify everything, so if you
are running a different distro and can verify this problem has been resolved, I'd appreciate hearing about
which specific version addresses it so I can publish the news here.
So why should you even care about this? You may have high core counts and running a kernel that has not yet been
patched. While this probably won't have any impact on any of your running applications - but do
you ever run top? iostat? sar? or any other monitoring tools? If you're reading this you probably run
collectl. Most monitoring tools are farily light-weight and for a good reason - if you're trying to measure
something you don't want the tool's overhead to get in the way. Unfortunately with this regression
it will now!
The Analysis
Whenever a monitoring change is made to collectl I measure its overhead, just to make
sure it's being done efficiently. The way I do this is simulate a day's worth of sampling at an
interval of 0 seconds and timing it. In this case I had recently added numa monitoring and initially found
this problem reading /sys/devices/system/node, but with more testing found it appears elsewhere
as well.
In the following example, you can see monitoring CPU data takes about 3 seconds to read almost
9K samples and write them to a file on a 2-socket/dual-core system. Very efficient!
time collectl -sc -i0 -c 8640 -f/tmp
real 0m2.879s
user 0m1.908s
sys 0m0.913s
Next I ran the same command on an 8-socket/48-core system and look at the difference. Note that the overhead
was so high and took so long I only took 1/10th the number of samples (I get impatient). This system is
running Fedora 14, which is a 2.6.35 kernel, and this alone is over 5 times the overhead
of the previous example which normalizing to a full day would be over 50 times the load:
time collectl -sc -i0 -c 864 -f/tmp
real 0m16.783s
user 0m3.003s
sys 0m13.523s
How can you tell if your system has this problem?
Before you panic, there are 2 things to keep in mind:
This only seems to effect systems with higher core counts, on the order of 32 or more. There
may be some impact at the 16 core level, but not very significant.
You must be running a kernel with the problematic code from the 2.6.32 release.
Since a simply uname command will tell you your kernel version, you might think that's all
it takes, but nothing is always that simple because most vendors patch their kernels and you can't
always be sure what code it's actually running.
One simple way to
tell for sure is to run the very simple test below which times a read of /proc/stat (which seems to be
the most heavily effected) by using strace see how much time is spent in the actually read.
As you can see the differences are dramatic! On my 4-core machine virtually 0 time is spent doing
the read while on the 48 core machine almost 15 msec is spent reading, and that's only reading
/proc/stat one time! Also remember - monitoring tools typically read a lot of different
/proc structures. Perhaps now you can get a better appreciation of how significant this problem
really is.
What is collectl doing about this?
While there is nothing collectl can do to reduce this overhead, staring with the next release it will
be including code upon startup times the reading of /proc on newer kernels with core counts of 32 or
more and warns the users if their systems exhibit this problem. It will also skip this test for RHEL 6.2
and SLES SP1. As I hear that other distros have been identified that have also fixed this problem,
I'll start excluding them from the tests as well.
This tutorial is intended to help you get started monitoring a system on which the
lustre filesystem has been installed. This is not
intended to be a tutorial on how to use such a filesystem. In terms of the basics,
there's really not much to say other then tell collectl to monitor lustre by
specifying an l with -s along with any other subsystems you may want to monitor.
As you should be aware of by now, collectl will try to display everything you've
selected in brief format if it can, writing all your data out on a single
line for each sampling interval.
This can get quite wide depending on how many subsystems you choose to monitor and
while you can cerainly specify collectl -s+l to add lustre to the default
subsystems, the output width is too cumbersome for this tutorial.
Therfore I'll use some other subsystems and switches in the examples
to mix things up, showing that there are a lot of possible combinations. In this
first example, we see what happens when you run collectl on a lustre client and
request cpu and memory data along with lustre.
Collectl is actually very intelligent about dealing with lustre because if you
were to run the identical command on an OST, it would recognize that too and
change what it shows accordingly as you can see below.
As expected, you can run collectl on a system that has any combination of MDS, OST and
client services and it will show you what you want to see. For more detail on some of
the more advanced concepts beyond the scope of this tutorial you can read more about
how collectl deals with Lustre here.
And finally, don't forget any of this data can be written to a file for continuous
logging, played back later and even converted to a format suitable for plotting.
In fact collectl is configured to monitor lustre by default when run as a daemon
so all you need to do to begin collecting the basic data shown above is
service collectl start.
Beyond the Basics
For many users of collectl, you are now sufficiently equipped to tackle most lustre
monitoring tasks, but for those more exotic situations there's so much more you can
do.
CLIENTS
Let's start off by looking more closely at client data. As with all other collectl
data, one can switch between summary and detail data by simply entering an
upper case L instead of a lower case one as I've done below.
The actual content of what is
displayed will depend on whether you are on a client, OSS (there is currently no
detail data for an MDS), but as you can see for clients, this data is broken down
by filesystem and just to make the display a little more interesting I decided to
show the timestamps in milli-seconds. Naturally if there is only one filesystem,
the data with match that displayed in summary mode.
Just to take it one step futher, it turns out that lustre actually tracks client I/O by
individual OSTs and sometimes that is more interesting, so there is a special option
for lustre clients, --lustopts O.
So what else can we look at? Lots! Lustre tracks readahead data which you can
show in brief format like this by using the R option noting this time we're
also requesting date/time stamps be included. Here we see the cache hits/misses
added to the brief display for the client.
Lustre also tracks client metadata, which you can request by specifying --lustopts M and
BRW stats which are selected by --lustopts B. However both are so wide they only
have a verbose form and the following shows both being displayed at the same time.
And if that's not enough, these even have a detailed mode of display and you can look
at them by filesystem or OST. You can specify any combinations of B,M and R with --lustopts
to show the associated data in both summary or detail formats, though only readahead
appears in the brief display. All others will force verbose format.
Object Storage Server
If you haven't figured it out yet, it's also possible to display OST level information
on an OSS by simply using the uppercase subsystem specification as you can see here,
noting just to be different I've chosen a second form of the date/timestamp.
You can also show BRW stats as both summary and detail data as well and since they
look identical to the way they're displayed for a client, I won't bother repeating
those forms here.
Metadata Server
As mentioned earlier, there is no detail data for an MDS nor are there any other
types of data other than that which can be displayed in summary mode.
Summary
As you have seen, there is a wealth of data here and often you may not even know what
you want to look for. When such is the case, just collect as much as you can and save
it in a file. Then you can play it back later and display it in multiple formats.
I just want to close with an example of a problem with readaheads and Lustre 1.4
which demonstrates the power of a tool
like collectl that can show multiple types of data at once because I know a lot of
people may still not be convinced. Consider the following sample which was collected while
doing random 32KB reads of a large file. See anything wrong? Do you know why the
network bandwidth is so much higher than the lustre client read rate? How long would it
have taken to even realize there is a problem?
It turned out the old algorithm triggered a readahead after 2 consecutive pages were read
and every 32KB read (which is 8 pages)
resulted in 1MB being read over the network, loaded in to cache and was then discarded!
If the value of max_read_ahead_mb is set to 0 for the associated filesystem on
each client involved, you see 2 things - the network rate drops to track the
client read rate and the misses goes to zero.
Caution - changing the value of the readable variable in /proc to 0 will eliminate
all readahead for all readers meaning any applications that do sequential
reads could have significant performance problems. In other words, this is not a
recommendation to turn off readahead but rather to be aware of what it does and if you do
choose to turn it off to be aware of the consequences
Note: This readahead algorithm has changed with V1.6 and lustre no longer triggers readahead
on the third page read but rather on the third sequential read.
There are actually 2 reasons for playing back a file, one being to generate
plottable files and the other is to simply examine
the data in the same format as you would see if running collectl interactively.
The following discussion applies to both cases, the only real difference is that to
generate plot files you include the switches -P and -f.
You tell collectl to play back one or more files using -p followed by any
combination of one or more filenames separated with commas or whitespace, noting you
need to quote the string if it contains spaces or wildcard characters.
The files will be played back as if a single file with monotonically
increasing sample numbers for each unique host. It should be noted
that if these files contain samples for different subsystems the resultant
stream will contain data elements for all, zero filling as appropriate. When
this occurs, a message will be displayed if -m has been specified.
Collectl will generate plot format if requested with -P,
writing the output to multiple output files if both summary and detail data is
specified or when a file with data for a differnet host or a different collection
date is encounted. NOTE: files that contain data that crosses
midnight will not force creation of a second file when the date changes.
Filtering with --from and --thru
You restrict the timeframe between which data is reported by using
--from and --thru. However, since collectl doesn't require you
to specify both switches nor does it require you to specify both a date and time
for each switch, it tries to make an intelligent guess as to what timeframe
you really meant. In most cases it guesses right but sometimes it guesses wrong.
The simple fix if it gets confused is to remove the ambiguity and just specify
full dates/times with both switches.
When you specify playback files using wildcarding in the name string, collectl
initially selects all the files that match. However, in some cases you may not
really intend for them to all be played back, especially if you selected a timeframe
using --from and --thru switches. Have no fear. Collectl will use
these switches to select the appropriate subset of files that match your selection.
Caution
When filtering wildcarded files using --from and/or --thru switches,
collectl compares those values with the timestamps of the filenames to determine
whether or not that file is likey to contain the data requested. However, collectl
doesn't know if a file contains data that spans midnight or not so it simply assumes it
doesn't since this is a rarely used feature. Therefore, collectl only
relaxes these tests when wildcards are not specified in the filename playback string.
Processing files that span midnight
The main reason for the complexity in the interpretation of --from and
--thru, is to allow collectl to deal with files that
contain data that crosses midnight.
As an example, consider a single file with data collected from midnite
on one day to 2AM, 26 hours later. If you want to tell collectl to process the entire file,
don't specify any time filters and it will report everything.
But if you want to process the date from midnight to 1AM,
you need to tell collectl which dates are involved! As stated in the rules above:
if you only specify --from 00:00-01:00 (note we're using the alternate switch
format), collectl will process 1 hour's worth of data
.
If you want 25 hours worth of data you will need to include the date of the second day
with --thru.
If you only want one hour's worth of data for the second day you will need to specify
both dates.
A word about the first record reported
Collectl always needs data from a base interval from which to begin
calculating changes in counters and that interval is never displayed.
In other words, if you collected data every 10 seconds starting at 10:00:00
and then played it back, the first time reported will be for 10:00:10.
In order to try and mitigate this when playing back data and specifying
a --from time, collectl attempts to read a sample from the previous interval so
that you actually see the time you requested.
Further, when mulitple files are processed collectl is smart enough to know if they are
contiguous to use the last set of data from one file as the base interval for the next one
and as a result there will be no holes in the data as reported.
However if they are not configuous a new base level must be taken for the new file and
its first record skipped. This can be confusing and probably not even that important
but consider 2 files generated contiguously:
If you process each file one at a time, their first samples will not be displayed
If you process them in one command using a wild card for the date/time, you will see
the first record of the second file
If you have 2 non-contiguous files you will see the same results whether you process them one
at a time or together using a wild card, that is no first record for either.
How --from and --thru are really interpretted
This section probably contains more detail than you should really care about and
is here more for completeness than anything else. Remember, these switches can
contain a date, a time or both! You can also use the shorthand form of combining
both switches under --from by separating the two with a hypen. In other
words you do something like: --from 12:00-13:00, noting you can use
an combination of dates/times on either side of the hyphen.
the default from and thru dates are 20000101 and 20380101 respectively
if no times are specified, each file will be processed from beginning to end
if no dates are specified, the time(s) apply to each file being processed. In
other words the switches --from 10:00 --thru 11:00 will result in data being
reported from 10:00 to 11:00 for each day! If only --from 10:00, data will be
played back every day starting at 10:00. If you want to the first day to start at 10:00
and subsequent days to be reported in full, specify a complete set of dates/times.
if one or both dates are specified, things become a little more complicated
if one date is specified, use the default for the one not specified
if both dates are specified, times only apply to the first/last dates
ignore any files created after the thru date/time
Important Tip
Perhaps the most important thing to keep in mind is that when you play back a file,
collectl will use the same switches as were specified during collection. In other
words if you collect cpu, disk and network data using -scdn, when you play
it back you will get cpu, disk and network summary data either displayed on
the terminal or written to a file. However, you could just have easily chosen a
different subsystem specification such as -scND in which case you'd still get
CPU summary data but now you'd get network and disk detail data. This feature can
be extremely useful especially when combined with different output formatting switches
such as -o and/or --verbose.
Introduction Collectl reports the standard values with respect to memory which are fully
documented under the data definitions. Trying to rationalize the
way these values relate to each other can be a frustrating experience because they rarely
add up to what you expect. Part of the reason for this that every byte is simply not
accounted for in every category. This can be further complicated because at boot time,
some devices will actually grab some memory that the kernel will never even see and so the
total memory will not always equal to the physical amount of installed memory.
If one looks at /proc/meminfo, which shows a lot more types of memory than collectl
reports, it begs the question "why not report it all?" and the simple answer is there is just too
much. Further, collectl uses a second file /proc/vmstat to gather virtual memory stats
which further adds to the volume of possibly candidates to report. Again, collectl tries to
report values of most use.
Brief, verbose and detail Like other data in these 3 categories memory also reports values in this way as well. However
there are a few important caveats to note:
Brief mode combines mapped and anonymous memory into a single value of Map
primarily as a mechanism to try and reduce the width of the memory columns since including too
many defeats the whole purpose of brief data reporting, which is to allow multiple types of
data on a single line.
Unlike other subsystems that report summary and detail data, memory summary data is not
the result of collecting detail data and adding it up. This means that unlike subsystems like
disks or CPUs in which you can record either summary or detail and still be able to play back
either, when playing black memory you will only get values for what you chose to record, otherwise
you will see values of all zeros.
There is no buffer or cache values reported as detail data because that information is not reported
Don't get fooled by cache memory vs used and free memory One of the most confusing things for people who aren't more familiar with linux memory management
is the interpretation of cache memory and its relationship to used and free memory.
As one might expect, as cache memory increases so does used and as expected free decreases.
What fools people is that the first (or many) times they see low free memory they think their system is
running out of memory when in fact is it not. If they reboot, the memory frees up, but then starts to fill again.
So what's going on? It turns out that whenever you read/write a file, unless you explicitly
tell linux not to, it passes the file through the cache and this will cause an increase in the amount
of cache memory used and a drop in free memory. What many people do not realize is, until that file is
deleted or the cache explicitly cleared, all files remain in cache and as a result if the system accesses
a lot of files cache will eventually fill up and reduce the amount of free memory.
Naturally linux can't allow the cache to grow unchecked, and so when it reaches a maximum set by kernel,
older entries will start to age out. In other words, reading a file will be extremely fast when in cache
but its access slowed to disk speeds when not.
The only real way to tell what is going on is to look
at the disk subsystem while accessing a file. If a complete or partial file is in cache, read I/O rates will
be much higher than normal. If a file is written that will completely fit in cache, again the I/O rates will
be very high because the rate at which cache is being filled is what is actually being reported. It is only
when a file is a lot larger than cache that the I/O rates slow down, operating only as fast as dirty
data in cache can be written to disk and is in fact the only real way to measure how fast your disk subsystem
actually is.
tip: --vmstat
Some people find the way vmstat reports virtual memory information to be very handy in
some cases. The only problem with vmstat is it doesn't write its output to a file and so even if
you wrap it in a script to write its output to a file you're now stuck with memory information
in a specific format and if you do want to plot it that takes a little more effort too.
Collectl's --vmstat switch is actually internally turned into --export vmstat and
so reports data the same way as vmstat does but now you get some added bonuses:
This data is always available when you record memory stats, you don't have to do anything
extra to make this information available. However note if you haven't recorded CPU stats those
fields will be reported as zero.
You can easily include the timestamp format of your choice, at least the 3 different ones
collectl allows and even include msec it you need that precision
You can playback data in vmstat format and then play it back again in other formats.
You can use this with colmux which now means you can do a
cluster-wide vmstat and show the top users sorted by the column of your choice
And of course you can still use any of colplot's memory plots
This is an excellent example of a very simple export module which can be a guide to help
you write others of your own choosing.
tip: don't forget about --grep
As previously mentioned, there is a lot more data contained in the memory /proc
structures than collectl reports. So does that mean you're out of luck if you want to
see the value of say Committed_AS?
Absolutely not, at least not during playback. Since collectl actually records the contents of
/proc data in its original formats in the raw files, you could always use linux's
grep (or zgrep) commands to search them for a particular pattern like this:
But unfortunately this does nothing to tell you what time the values correspond to. You could
have included the reporting of fields with >>> in them and you'll see the UTC timestamps, but
those aren't easily mapped to conventional time formats. Now look at this:
Pretty slick! And since this is collectl/playback, you can use other switches like
from/thru or even change the timestamp format and/or see msec too. Also remember
this trick can be applied to any data collectl records, though memory tends to be the most
interesting.
The answer to this question is fundamental to understanding system monitoring in general,
whether you're using collectl or some other utilities. To really understand what your system is
actually doing you should be looking at individual disks, cpus or networks. But in reality that's
often too much to keep track of unless of course you have a single-cpu/disk system.
Let's say you have multiple devices such as disks and the one you're interested in is
misbehaving and reading or writing too slowly. Won't the total disk activity also be low? Similary if
disk traffic is being reported high on a lightly loaded system won't this also jump out you by simply
looking at the total disk activity? The same can be said about networks and most other subsystems for
which there is summary data and simply looking at the totals will often alert you to the fact that
something is not right. The key thing to keep in mind that you are looking at totals.
CPU monitoring can be a little tricker as these are reported as averages as opposed to totals
and as the number of cores increase so does the divisor of the calculation. In most cases when you
have a system with excessive load, it will effect all CPUs and so will be very visible even as an average,
but in some cases it won't. What if you have a 2 core system and see
a CPU load of 45% when you're expecting a much lighter load? Looking at individual CPUs you may see
one running at near zero load and the second at 90%. Your only clue was that the 45% load was unexpected
and so you looked closer. But what if you had a heavily loaded CPU on a 48 core system? You'd never
even realize it. In other words, just pay attention.
Stated slightly differently, summary data is often a starting point to help identify potential trouble ares
and from there you can determine if you need to dig deeper.
So why brief data?
It you have ever tried to look at multiple lines of different text and identify what was changing over
time you should already know the answer - it's really difficult! For example, here's what collectl might
show for CPU, Disk and Network data:
and that's only 2 samples! Think about trying to watch level of detail changing every second and identifying
changes? Extremely difficult if not impossible. It might be a little easier to watch in top format which you
can get by including --home, but now you lose the previous records to compare the values to.
Now consider the fact that in many cases seeing network errors or disk merges or even the percentage of time
the CPU spent processing interrupts, while important, may not be when trying to identify anomalous behaviors.
And that's where brief mode comes in. Here we are identifying those few nuggets of information which will
tell us whether or not things are functioning as expected such that we can display them all on the same line
and make it easier to spot change. In fact, during the following run I did a ping -f and see how easy it is
to spot the network burst?
Now, if you think there is a network problem you can then run collectl in verbose or detail
mode and only look at network and not be distracted by other data.
In summary, just keep in mind that there is no single recipe for how to monitor a system, what format to
display the output in and how to drill deeper. However, as you become more familiar with the types of data
and collectl formats your ability to better utilize collectl will increase.
The following table needs a little explaining as it is rarely the case that the output of a
particular utility maps identically to an equivalent collectl command. In some cases
a single sar switch will approximate a single collectl subsystem and in other case it
takes multiple sar switches to do so.
The point is this matrix is only a rough guideline to some command equivalences and
simply changing the case of the collectl subsystem or adding the verbose switch will
provide even more output. There are even some options (see the uppercase O
switch) that can also effect what gets displayed.
And of course there is the lowercase o
switch which allows you to specify formatting options like date, time, etc.
In other words there are simply too many permutations to list them all here and you're
just going to have to experiment.
Finally, there is a special case worth noting. There is no collectl command equivilent
to sar -R. It's not entirely clear to me if this data is needed, but it
certainly could be added if the demand is there.
Collectl can generate output in either summary or detail format. Furthermore,
summary data can be generated in either brief or verbose format. Although
verbose is the most complete, the brief format, which is also the default, is the one viewed by
most users.
The most important thing you should know about Infiniband monitoring is that it
is destructive. What is meant by this is that every time
collectl reads the counters from the HCA it immediately resets them to zero, thereby
destroying their previous contents. You should also note this does
not apply error counters, which are never reset.
The obvious question is why? and perhaps the less than obvious answer is
because when the hardware specifications were written for the Infiniband HCAs it
was decided that performance counters would not wrap, probably because nobody
thought someone might want to do continuous sampling. In any event, at even modest
traffic rates HCAs with 32-bit counters quickly reach their maximum values and
stop incrementing, rendering them useless for performance monitors like collectl.
Collectl's solution to this problem is to read the
counters and immediately reset them to 0. As long as the next sampling period occurs
before the counters fill up, this methodology comes reasonably close to
reflecting the traffic rates (some counts are lost between the read and reset).
However, this methodology has a downside in that while collectl is monitoring
the Infiniband stats, nobody else can (including other copies of collectl).
Unfortunately there is no solution to this problem short of redesigning the HCA
and that's simply not going to happen. A second alternative would be to come
up with a mechanism in which the read/rest of the counters are moved into an OFED
module which exports these to /proc or /sys as rolling counters. This was in
fact done in a pre-ofed version of Voltaire's IB stack which is currently supported
by collectl. If someone would like to hear more details on how this was done, feel
free to contact me or to post something in a collectl
forum or to the
mailing list.
If you want to run collectl but also prevent it from doing destructive monitoring,
simple comment out the line in /etc/collectl.conf that begins with
PQuery = and you will be informed by collectl that Infiniband monitoring
has been disabled whenever someone try to monitor it.
Monitoring Mechanics
The main purpose of this section is to help you understand how monitoring works so when
it doesn't you might be able to figure out what went wrong. There are 2 different ways
collectl can monitor Infiniband, one for the OFED stack, which is the Infiniband Stack
of choice these days and the other for pre-OFED.
OFED
The OFED stack can be identified by the presence of the /sys/class/infiniband
directory. If there, collectl looks inside to find which HCAs are present and which
ports are active. This information is then used to query the HCA via the perfquery
utility.
Unfortunately, with each release of OFED that utility seems to move to another location
and collectl tries to react by using a search path in /etc/collectl.conf. As of the 2.5.1
release of collectl, if it still can't find the utility it will try to find its location
with rpm and then add its path to collectl.conf. If a future OFED release eliminates or
replaces perfquery collectl will break.
Pre-OFED
If it is determined the OFED stack is not running, collectl will try to use the
get_pcounter utility to query the HCA, which is in fact what
perfquery is based on. Since this utility has also moved around during
various releases, it too has a search list in collectl.conf, but unlike perfquery
if it can't be found no additional heroics will be attempted to find it. If you
don't have this utility on your system (and even if you do) you should probably
consider moving to OFED.
Debugging
Collectl has a variety of debugging capabilities built into it, the main one being
the debug switch -d. To use this switch you specify a bit mask which is then applied
against a variety settings which tells collectl what to display. For
debugging interconnect problems simply use -d2. All possible bit settings
and their meanings are listed in the beginning of collectl itself.
If collectl runs without errors but you're not seeing IB traffic being reported
when you think you should, you can always use -d4 or even -d6, which show
the values of the counters returned by both perfquery and get_pcounter. If they
don't change something outside of collectl must be wrong.
One example of a non-collectl problem was a system had IB configured and started
which could be verified by seein an ib0 interface show up with ifconfig.
However, when
running collectl -sN, which will show the traffic over all the network
interfaces, there was never any traffic on the ib interface however there
was unexpected traffic on one of the eth interfaces. Clearly something
was wrong and looking at the routing showed the routes were set such that all
traffic to the infiniband address was being routed over the eth interface.
Prior to V2.5.0, collectl reported the total number of interrupts across all CPUs
as part of the CPU summary data and nothing about interrupts in the CPU detail.
Since interrupt counts are actually reported in the kernel for each CPU
by individual interrupt number, that information will now be made available both
in summary and detail formats.
However, a slightly different methodology for categorization will be used because
interrupt summary data will be reported by CPU and detailed interrupt data will
break out the data by individual interrupts for each CPU.
Requesting Interrupt Statistics
Since -si has already been taken for reporting inode statistics,
interrupt summary data should be requested using -sj and detail data using -sJ.
Interrupts are also treated differently than other statistics in a couple of ways:
Rather than reporting a fixed number of fields like other summary data, the number
of fields are variable and equal to the number of CPUs
If one specifies -sCj one will NOT get a separate verbose format for the Interrupt
Summary, but rather will have those details included as part of the CPU detail report
as shown in the examples below.
Plot Data and s-expressions
Although looking at interrupt data in brief or verbose fits quite well with
collectl's reporting methodology, it doesn't fit the plot and s-expression
formats. This is because both expect a fixed number of fields in for summary data and
a variable number of fields for detail data, indexed by a device number. However,
interrupt summary data is actually variable size based on the number of CPUs and the
detail data doesn't really fit with anything and so trying to do so will generate an error.
Therefore to report interrupt data in plot or s-sexpressions
you will be required to request CPU detail data as well since interrupts are reported
as part of that data.
Examples
The following examples only show interrupt data except where CPU data is required. Any of
these reports can be extended to include other data types.
This first example shows the basic interrupt output displayed in brief mode on an 8
processor system, with timestamps. If you include --verbose you essentially get the same
display except with more significant digits for each interrupt.
As mentioned above, specifying -sCj is special as it takes the data for 2 separate
subsystems (CPU detail and interrupt summary) and combines them in a single display as
shown for two sampling periods on a 4 processor system:
A third form that can be particularly useful is the Interrupt Details which shows
the distribution of the individual interrupts across the CPUs. What makes this form
especially handy is it only shows those interrupts that changed during the
monitoring cycle which is considerable easier to read than /proc/interrupts itself.
If the number of CPUs change during processing, this can only be detected when monitoring CPU data
at the same time. If you are only monitoring interrupt data and there is a state change things will
get very messy. As users typically monitor Interrupts and CPU data at the same time it is not
felt to be worth the extra effort or processing overhead to try and accommodate this rare case.
In version 2.6.22 of the Linux kernel, the slab allocator has
been replaced by a new one called SLUB, the Unqueued Slab Allocator,
and more importantly from collect's perspective,
the way slab statistics are reported has changed as well.
Rather than reporting all slab data in the single file /proc/slabino,
there is now one subdirectory for each slab under /sys/slab.
But before getting into all that here's a quick review of how slabs are
organized by referring to the following diagram:
As you can see, for a given slab name there are multiple slabs and each slab
consists of multiple objects. When a process requests an allocation of slab
memory it is provided as an object from a slab if there is one available. If
there are none, a new slab is allocated and the object provided from it.
Furthermore, slub allows slabs of different names
but whose objects are the same sizes to share the same slab as you can see below
for the slab with the very ugly name of :0001024, which in this case is a
slab which contains 1K objects. These additional
entries are called aliases for obvious reasons:
Good news! The slab memory field reported in /proc/meminfo finally matches the total
memory reported the individual slabs and so the need for collectl's slab summary
usage for all slabs has been reduced. However, when selecting a subset of
slabs by filter(s), the summary will show the totals for the selected slabs
and will therefore be more useful. More on this in the examples below.
The main pieces of information collectl reports on for each named slab are the number of
slabs that have been allocated, the corresponding number of objects and the number
of objects that have actully been allocated to processes. Collectl reports the total
memory associated with the slabs as well as the amount of slab memory actually
being used by processes. It also reports some constants such as the number
of objects/slab and the physical sizes of both objects and slabs.
Examples
The following examples show several different output formats and the commands
used to produce them. One should note there is similar output for the old style
slab data but. It should also be noted that an interval of 1 second
has been chosen in each case and one should always consult the help
and/or man pages for more detail as there are many other formatting options.
Perhaps the easiest way to get started it to just type the command
collectl -i:1 -sY and later add some additional switches to see their impact.
For those new to collectl, you should also realize that collectl can run as a
daemon logging all this in the background for later playback and that all the
different subsystems it supports can be include by simply adding them to -sY.
Summary
This is the verbose, time-stamped slab summary output for
only those slabs beginning with 'blk' or 'ext3'
collectl -i:1 -sy --slabfilt blk,ext --verbose -oT
# SLAB SUMMARY
# <---Objects---><-Slabs-><-----memory----->
# In Use Avail Number Used TotalK
13:21:10 120625 124233 30701 113894K 122832K
13:21:11 120625 124233 30701 113894K 122832K
13:21:12 120625 124233 30701 113894K 122832K
Standard Detail
Here's the same report, only now we're looking at details and tossing in msec
timestamps
Here we see the same output again only this time we're simultaneously
writing a large file and choosing to report on only those slabs which
have changed between monitoring intervals. To make the output a little
more interesting we've added filtering on 'dentry' as well:
Version 3.1.1 of collectl introduces a new format for slab data as specified by --top and produces
output in a format similar to the slabtop command, but adds two new fields TotChg and
TotPct, which allow one to see and sort on the change to the actual physical memory allocation.
These two new fields allow one to see which slabs are changing the most and/or having the most impact on
physical memory because sometimes a small percentage change to a very large slab can make a big difference
in memory while a large percentage change to a small slab may not and hence to 2 additional sorting alternatives.
As with process data, the argument of the switch
describes how to sort the data and an optional line count. In the case of slabs the sort name matches the
column headers making them easy to identify and if you forget how to get started, just run collectl with
--showtopopts. Naturally filtering can be applied as well and the following shows the output when
used with a couple of filters.
Similar to process data analysis, the --slabanalyze switch will cause the slab data to be analyzed and
summaried in its own file with the .slbs, written into the directory pointed to by -f. The focus of
that analysis is currently on how much memory is actually consumed by each slab, the contents of that file
making it possible to determine which slabs may have had the greatest impact on memory utilization. Slabs which
do not allocate any memory will not be included. The following is an example report:
This report shows the starting/ending values of memory utilization for each slab as well as the minimum
and maximum values over the course of the day. The final two, and perhaps the most interesting
columns, show the difference in memory usage over the course of the day as well as the percentage
change over the low value.
As with the process analysis report, you can control the timeframe of the report using --from/--thru
and unless you explicitly specify one or more subsystems no other data will be reported. On the other hand if you
do specify -s, you will get the additional data reported in the standard way in plot-formatted files.
Slabs and the rawp file
As of collectl V3.3.5, it is now possible to request slab data as well as process data be logged to a
separate file as specified by the -G and --group switches. To read more about the mechanics of this
see Logging and in particular the section on Grouping data into 2 files.
If you have indeed chosen to use this mechansim there are a few things that have changed in collectl's behavior:
If you play back the pair of files as a set the right things will happen. Collectl will play back the non-slab
containing file as if slab data was never requested, defaulting to brief format unless something was requested
that explicitly forces verbose format, and then play back the other file as if only slab (and possibly
process) data had been requested. In any event, the output will not be interleaved as each file is played
back independent of the other.
If you explicitly play back the file with the slab data and specify subsystems other than slabs or
processes, they will be ignored in the output. If you play back the file without the
slab or process data and specify slab or process detail, they will also be ignored.
If you do not use -G and write slab data to the same file, -sy will automatically put you in
verbose mode as brief slab data is no longer reported. If you want to see brief slab usage, use -sm.
All raw data is recorded in a single compressed file with the extension
raw.gz if the perl compression library has been installed. If that
library is not there it will be written to a non-compressed file with the
extension raw. The only exception to this rule is the process
raw file which can be useful on systems with a large number of processes
(see the description of --tworaw) and which has the extension rawp.
Plot Data
There are actually 2 main types of plot data - summary and detail.
Summary plot data, for those subsystems selected with lower case letters,
is always stored in a single file, one line per time period,
with the extension tab. The primary reason for this is that the data for
each subsystem is of a fixed length and there is really no benefit in
separating it into multiple files.
Detail plot data, which is typically for devices for which there can be multiple
instances such as CPUs, is recorded in one file per detail type with an extension
that reflects the type of data stored in that file.
Each line contains instance data of a fixed number of
fields for that particular device. Although TCP do not have instance data,
it does have a detail component and is also written to its own detail file.
Process and Slab data are also treated like detail data because they too require
multiple lines per monitoring period.
Exception Data
Exception data is written to a file in the same format as detail data with an
X appended to its name. Since exception data is not of a known format
across the entire device as is detail data, it cannot be written as a single
line, but rather is written as one line per device. Each line is prefaced with
a date/time stamp and the number of the device (0 based).
Collectl Messages
Periodically collectl logs various types of messages to its own message log to avoid
situations in which an unexpected situation or a collectl bug might causes the
flooding of /var/log/mesasages. However, the more serious messages are written to
both as described here.
This format does NOT include data for any individual devices such as cpu, disk,
network, nfs, lustre, process, slabs or tcp. If you do select any one of them
collectl will force --verbose format.
Buddy (Memory Fragmentation) Data, collectl -sB
The smaller the value the more accuracy as follows:
0-9 are represented by the digits 0 through 9
10-99 by the letters a through i
100-999 by the letters j through r
>1000 by powers of 10, using s through z
Remember, the whole purpose of monitoring fragments in brief mode is to identify trends,
particularly when there are a small number of them.
#<---Memory-->
# Fragments
smkj9576040
CPU Data, collectl -sc
#<--------CPU-------->
#cpu sys inter ctxsw
cpu
Percent of time the cpu was busy during the current interval averaged
across all CPUs and is actually the total percentage of time the CPU in one of the following:
system, user, nice, irq, soft-irq and steal. Note that this does NOT include time
spend in I/O wait.
sys
Percentage of time the cpu was executing in system mode during the current
interval. This includes all those modes as above except user and nice to to
determine the amount of time spent as a user you need to subtract these from the total cpu
field.
inter
Total number of interrupts/sec.
ctxsw
Total number of context switches/sec.
Disk, collectl -sd
If you specify filtering with --dskfilt, the disks that match the pattern(s)
will either be included or excluded from the the summary data. However, the data will
still be collected so if recorded to a file can later be viewed.
Average incoming packet size in KB. This field is only
included if --xopts i or --iosize included
KBOut
KB sent/sec.
PktOut
Packets sent/sec.
Size
Average outgoing packet size in KB. This field is only
included if --xopts i or --iosize included
Errs
Count of current errors. Since these
are typically infrequent, it is felt that reporting them as a rate would result
in either not seeing them OR round-off hiding their values.
Inodes/Filesystem, collectl -si
#<----Files--->
#Handle Inodes
5100 116110
Handles
Number of allocated file handles
Inodes
Number of inodes in use
Lustre
Lustre data actually falls into one of 3 categories - client, mds and oss. Collectl
determines the type of system it is running on (a system can have multiple
personalities) and reports on all it finds, unless specifically selected via -L.
Average read size in KB. This field only included
if --iosize specified
not necessarily from the lustre storage servers.
KBWrite
KB Writes/sec delivered to the storage servers.
Writes
Writes/sec delivered to the storage servers.
Size
Average write size in KB. This field only included
if --iosize specified
The following format of lustre client data is selected by including -OR and
adds readahead statistics to the previous six, noting the Size fields
are dependent on --iosize being specificed.
Total free memory, which unfortunately is NOT the
difference between total memory and the following amounts allocated to used memory.
buff
Memory used as system buffers.
cach
This is also commonly known as the file system buffer
cache as buffered I/O uses this memory to cache the data.
inac
Inactive memory.
slab
Total memory allocated to slabs.
map
Total mapped memory, which include AnonPages.
Note
If you include --memopts R, these values wil be displayed as changes/sec between intervals
rather than absolute values. This switch will also honor -on in that the values will not
be normalized to a rate but rather displayed as changes in size per interval.
Network, collectl -sn
If you specify filtering with --netfilt, the names that match the pattern(s)
will either be included or excluded from the the summary data. However, the data will
still be collected so if recorded to a file can later be viewed.
Also be sure to keep in mind that like all other data, network counters are always normalized
to /sec rates. This means that something like errors, which themselves might be really
small, could be reported as 0 if less than 1/2 the sampling interval. To see unnormalized
values use -on.
KB received/sec over all real network interfaces and therefore
excludes 'lo' and 'sit'.
PktIn
Packets received/sec over all real network interfaces.
Size
Average incoming packet size in bytes. This field is only included
if --netopts i or --iosize specified
KBOut
KB sent/sec over all real network interfaces.
PktOut
Packets sent/sec over all real network interfaces.
Size
Average outgoing packet size in bytes. This field is only included
if --netopts i or --iosize specified
Error
Total incoming/outgoing errors/sec. To see individual error counts, use --verbose.
This field is only included if --netopts e specified
NFS, collectl -sf
As of version 3.2.1 collectl now collects all types of nfs data, both clients and servers as well
as versions 2, 3 and 4. In brief format it therefore reports summaried data across all nfs types
as shown below. If --nfsfilt was included in the command to limit the types of data
reported, those values will be included in the header line as a reminder as shown in the second
form to the right, in which case only V3 server and V4 client data are being summarized.
#<------NFS Totals------> #<------NFS [s3,c4]----->
# Reads Writes Meta Comm # Reads Writes Meta Comm
Reads
Total nfs reads/sec.
Writes
Total nfs writes/sec.
Meta
Total nfs meta data calls/sec, where meta data is considered to be
any of: lookup, access, getattr, setattr, readdir and readdirplus, noting that
not all types of nfs version report all as V3 clients/servers do.
Comm
Total nfs commits/sec.
Slabs, collectl -sy
#<----slab---->
# Alloc Bytes
Alloc
Total Number of slabs allocated
Bytes
Total Number of bytes allocated as slabs
Sockets, collectl -ss
#<------Sockets----->
# Tcp Udp Raw Frag
Tcp
Total TCP sockets currently in use.
Udp
Total UDP sockets currently in use.
Raw
Total RAW sockets currently in use.
Frag
Total number of IP fragments queues currently in use.
TCP, collectl -st
The TCP data collected and actually depends on the value of --tcpfilt, which by default is
set to itcu, which stands for IP, Tcp, Udp and ICMP. A 5th filter T
will result in Tcp Extended values being reported.
Here we see a few things including a burst of cpu activity when the test first
starts as well as an I/O rate of about 30MB/sec which corresponds to what dt is
telling us in the following summary line:
Average transfer rates: 32051995 bytes/sec, 31300.776 Kbytes/sec
If we compare the write rates to the number of writes we can also infer writes
of about 128KB which is good to know because that means we're being efficient
in the size of the data blocks being handed to the driver. However if we don't
mind using the extra columns, we can include --iosize, which tells
collectl to include the average I/O size when using this default display format
also known as brief mode. In verbose mode the I/O sizes are always included.
This may also be a good time to mention screen real estate. There is a lot of information
that collectl can display and everything takes space! More often than not you don't
really care about time and so by default it isn't displayed. However there may be times
you do care and so you can simply add the switch -oT add the option of time to the display.
In fact, sometimes you may want to include the date as well in which case -oD will do
both. You can even show the times in msec by including m with -o, which can be
useful when running at sub-second monitoring levels and/or if you want to correlate data
to system or application logs with may themselves have finer grained time.
Here's an example of the command collectl -scd -i.25 -oDm
which shows the cpu and disk loads every quarter second and includes the date and time in msecs:
So what about that CPU load? Given that this is a 2 CPU system we might be
interested in seeing how that load is being distributed by running the command
collectl -sC, since an uppercase subsystem type, like cpu, disk
or network tells collectl to show instance level details:
Now we see a big burst of CPU load and not much from disk. Furthermore dt
is reporting
Average transfer rates: 872960833 bytes/sec, 852500.813 Kbytes/sec
which in fact confirms that reads are coming from cache and not disk since
no local disk can read at this rate! In general, when doing disk I/O testing one
should use file sizes that are larger than cache to force all I/O to come
from disk. So repeating the tests with a larger file we now see more
realistic read rates:
So just what is happening to cache during testing? To see memory utilization
we can simply add the memory subsystem to the default selections as
collectl -s+m but that also makes the display wider and since for our
purposes we don't need network information I'm just going to run the following
collectl -scmd:
and watch the cache fill up. In fact, if we keep running collectl eventually
we use up all available memory (but that's what it's there for) and even after
the test completes and there is no more I/O, we still see hardly any free
memory. But that too is ok because until someone else needs it or deletes
the file, that data stays in cache. Look at the last sample where I manually
deleted the file. You can see the cache drop to 204M and the free memory
rise to 2G during a single reporting interval:
I first see a batch of over 3K nfs writes which also include 6 metadata calls,
which are clearly doing a variety of directory accesses to see if the file
currently exists as well as for creating the new one. During the next
interval the network starts sending the bulk of the data over the network,
which also include 4 commits (nfs does commits for a batch of writes as
opposed to a single commit/write which would be excessive and slow). In
the intervals that follow, nfs need do no more writes as they've already been
queued up and so for the next several intervals all we see is the network
traffic. The CPU load has also gone down because the data has already been
moved into the outbound I/O buffers. For more details on nfs, see
this page.
So in conclusion you can see there is really quite a lot you can do with just a few basic
switches and I haven't even gotten into --verbose, which as they say is an exercise left for
the student. So try some simple dt tests yourself or use you own personal favorite
load generator, while trying out collectl -sc --verbose or collectl -sm --verbose or even
collectl -sn --verbose. You can even put them all together as
collectl -scmn --verbose, but then as you'll see you end up using a lot of that valuable
screen real estate. As a final bonus, try adding the --home switch which move the cursor to the
home (upper left-hand corner) position of the screen. Think of this as something like the
linux top command (collectl also has a --top switch for displaying slab/process data)
since each sample is displayed at the top of the screen. That command
would then look like collectl -smcn --verbose --home.
There are far too many combinations of switches and output formats so only a few of the
more basic ones will be shown below. These examples show both the command and in most
cases the resultant output. For more examples see both the
FAQ and collectl man page
after you install it.
Interactive Commands
The following examples show the results of running collectl interactively and seeing
system performance numbers in real-time.
Default
Notice that in this mode you see one line per sampling interval. You are only limited
by the width of your terminal window.
In this mode you give up one line per interval and are rewarded with more details than
could fit on a single line. This format always includes the date and time.
If you would rather see details on specific instances, use the uppercase subsystem
names with -s, so rather than the default of -scdn use -sCDN,
which also forces
--verbose, noting that you can also mix lower and uppercase subsystem types.
[root@poker]# collectl -scft -oT
waiting for 1 second sample...
# <--------CPU--------><------------TCP-------------><------NFS Totals------>
#Time cpu sys inter ctxsw PureAcks HPAcks Loss FTrans read write meta comm
08:12:20 0 0 1007 120 1 0 0 0 0 0 0 0
08:12:21 1 1 1077 400 1 0 0 0 0 0 0 0
When you just don't know...
A great way to familiarize yourself with the types of data collectl can
generate is to do (as of Version 2.6.0) --all --verbose and
see all the summary data generated at once,
excluding processes and slabs. Since this can be a lot to watch,
especially as it scrolls off the screen between samples, the --home
can be your friend. It will clear the
screen between samples and remove extra lines to give an appearance
of a continuously refreshing screen-based utility. Try it with different
combinations of subsystems to reduce the amount of information
displayed.
This mode is often used when running a test of a limited duration from a couple of minutes
to several hours or more and since collectl is not being run as a daemon, the default
sampling rate is 1 second.
Collect 100 Samples and Exit
[root@poker]# collectl -c100 -f/tmp
Run Until Terminated With ^C
[root@poker]# collectl -f/tmp
Playback Mode
There are a couple of things to remember about playback:
You need a data file. When run as a deamon via service collectl start
data will be written to /var/log/collectl by default.
You can playback a subset of data by inclusion of the -s switch
Just about any formatting switches you can use interactively you can use during playback
You can select a subset of the time period with --from and --thru
Playback Between 2 Time Periods
In this example we're not selecting any device details and so the output defaults
to brief mode and all data is printing on the same line. We've also chosen to
display time in msecs and use --from to specify both times.
Notice here that we're using -oD instead of -T and so both the date and time are displayed.
We could have also chosen to force the time in msecs but chose not to in order to save
screen real estate.
collectl-3.6.9/docs/Gexpr.html 0000664 0001750 0001750 00000025164 12230241727 014413 0 ustar mjs mjs
Exporting Data In Ganglia Format
Exporting Data In Ganglia Format
Introduction
With the release of Collectl Version 3.3.1, one can now send collectl data directly to
a ganglia gmond in binary format using the custom export gexpr.ph. This results
in several benefits for existing users of ganglia:
You can now get the benefit of collecting all the data collectl can generate
at a very low overheard.
Since all collectl instances send this data at the same time, system noise on
clusters running fine-grained parallel jobs is reduced.
You can still log all the data collectl collects locally and only send a subset
to gmond, making it possible to collect far more data than ganglia can handle without
overwhelming the gmetad. This is especially important on large clusters. In fact, if
you choose you can have collectl only send ganglia the core data it currently provide
graphs for.
And finally, you can monitor as frequently as you like and send data to ganglia at a
coarser frequency.
As of V3.5 of collectl, the experimental status of this capability has been lifted
since enough people are currently using it to verify it works as described.
Ganglia Configuration
The following figure shows one possible configuration of a large cluster running ganglia and
is only intended to be illustrative. For complete documentation on how to set up and configure
ganglia see the official wiki.
This configuration assumes one has set up ganglia gmonds in a hierarchy, such that those at
the bottom of the tree collect system statistics and send them up to a higher level
aggregation gmond and ultimately percolate up to the gmetad which writes the data to
a round-robin database.
To use collectl as a data source there are 2 alternatives. In the diagram below at the left
you simply have collectl send data to a local gmond to supplement whatever data it is
already collecting, noting there won't be any way to record any of gmond's data
locally. The diagram at the right would replace all gmonds and have collectl do all the data
collection, optionall logging data locally, and sending data to an aggregation level gmond.
Whatever method you chose you must ensure the gmond(s) are listening on their udp receive
channel and enable the ganglia communications feature in collectl as described
in the following sections. There are probably other hybrid
configurations which are beyond the scope of this document as well as this author.
One last component is configuring the rrd to use the metrics being supplied by collectl and
the details of that discussion are beyond the scope of this document as well.
Usage
Like any other custom export used by collectl, you tell collectl to
use gexpr with --export gexpr and in addition include the gmond hostname:port
followed by one or more of the standard switches.
There are also 2 more switched unique to gexpr and they are:
g - only send core variables to gmond, using the standard names gmond
has predefined graphs for (see the second verification example below).
G - send core variables to gmond, using the standard names gmond
has predefined graphs for, but also include any additional variables collected
by collectl and possibly filtered by gexpr's s switch
The following example shows collectl gathering data on many system components but only sending
cpu, disk, memory and network data to a gmond on system gmond using port 8108.
It also sends a set of data every 20 seconds while writing the data to a file in plot
format to a local file in /tmp every 5 seconds.
This module also supports sending its output to a multicast address, which is used if
hostname in an address in the range of 225.0.0.0 through 239.255.255.255. To use this
feature you will have to first install
IO::Socket::Multicast which in turn requires the module
IO::Socket::Interface
be installed as well. Note that both these modules may be updated and so you should verify
you're actually installing the latest one.
There are 2 ways to make sure everything is working as expected, the first is to make sure you're
using the ganglia export module correctly and to do this you can use the debugging parameter.
Like collectl itself, which has its own debugging variable (see d), the value of this
debugging variable should be interpreted as setting a bit mask, where each bit results in
different behavior. Refer to the header of gexpr for the complete set, noting
the following example uses only 1 bit, namely the one for printing the data being set over the socket.
If for some reason you want to simultaneously disable actually sending the data over the socket
to ganglia use the debugging value of 8 or a combined value of 8. Note that in this latter case
you are still required to supply a socket:port because they will be opened but also realize
you can use just about anything you want since no data is actually sent over it.
This second example shows the use of the g option which only sends the
core ganglia data using ganglia variable naming. Also notice a nifty
trick that since we're telling gexpr not to send the data over a socket, we
can use a dummy network address/port and same ourselves some typing. Also
note that since we didn't select memory or network stats, they aren't sent
to ganglia either.
Assuming you're now successfully calling gexpr and can see the output above, it's time to
make sure it is going to the expected gmond. The easiest way to do this is to simply run
the gmond with debugging enabled at a level of at least 2
and make sure it can see the data from collectl, being sure you haven't set a debug level
of 8 in gexpr. If it can see data in gmond (note that only the variable names and not their
values are reported by gmond), you're done. If not, you need to make sure gmond is
correctly listening for udp data and that the port it expects to see data on is in fact the
one gexpr is sending to. If all looks correct, it may be necessary to watch the network
traffic with a tool like udpdump or wireshark.
This is an example of what you should expect to see, noting that in this case gmond
is still configured to collect its standard metrics, which can be disabled
in gmond.conf since you no longer need these.
gmond -d 2
loaded module: core_metrics
loaded module: cpu_module
loaded module: disk_module
loaded module: load_module
loaded module: mem_module
loaded module: net_module
loaded module: proc_module
loaded module: sys_module
udp_recv_channel mcast_join=NULL mcast_if=NULL port=8108 bind=NULL
tcp_accept_channel bind=NULL port=8109
Processing a metric metadata message from cag-dl585-02.cag
***Allocating metadata packet for host--cag-dl585-02.cag-- and metric --cputotals.user-- ****
saving metadata for metric: cputotals.user host: cag-dl585-02.cag
Processing a metric value message from cag-dl585-02.cag
***Allocating value packet for host--cag-dl585-02.cag-- and metric --cputotals.user-- ****
saving metadata for metric: cputotals.nice host: cag-dl585-02.cag
The name of this subsystem is a slight misnomer because in addition to monitoring inodes it
also monitors dentries as well. The information being reported is quite self-explanatory
and there should really be no mystery here.
However, it may be worth noting that while inspired by sar -v the information isn't
quite the same in case that's what you're expecting. Sar also reports on the number
pseudo terminals and somehow that didn't seem to fit with the types of data collectl reports
and so was left out.
Another difference worth noting is that while both report the number of open files, actually
calling them allocated file handles, it was felt that this number all by itself wasn't providing
as much info as it could and so the percentage of the total has also been included in collectl's
output.
While I wanted to do the same with dentries, reporting the percentage of the max, I ran into
a problem - the number of unused cache entries is wrong, or at least the definition is!
Rather than decrease as new directories are created it goes up, which makes no sense to me and
so trying to report it as a percentage doesn't seem to make any sense. However, a closer
investigation makes me think this is in fact tracking the slab objects named dentry_cache
which does increase as new directories are created. However something still isn't quite right
because while close, the numbers aren't close enough.
Investigation will continue, though at a lower priority.
If/when resolved it will be fixed/documented accordingly.
Depending on which combination of switches are selected, collectl will run in
one of 3 main modes with various options for added flexibility.
The most basic mode, which you get if you don't select one of the other 2,
is display. In this mode the output is displayed on the terminal
in real-time as it is collected. In record
mode, specified by the -f switch, data is written in real-time to a
directory of the user's choosing with an optional prefix. In playback mode, selected with -p,
data is read from a file that was generated in record mode at an earlier
time.
The format of the results can also be selected as either Terminal or
Plot. Terminal data is always displayed on the terminal while
Plot data, selected by including
-P with any of the 3 modes, can be either written to a file or displayed on
the terminal. Since plot data is not intended for human consumption, the
reason one would typically send it to a terminal would be with the intent of
redirecting the output to a file or piping it into another script.
Using the -f, -p and -P switches in different combinations result in the
following behaviors:
No switches
Data is displayed on the terminal as formatted text
-P
Data is displayed on the terminal in Plot Format
-f file
Raw data is written to the file (whose name is constructed
by collectl) in the same format as it occurred in /proc, with the extension raw.
For more details on file naming see file naming.
-f file -P
Data is written to the specified file in plot format,
with one or more of a number of extensions depending on what detail data may
have been requested.
-p file
Data is played back from the raw file specified by -p and
displayed on the terminal as formatted text. If one wishes to view a subset of the data
recorded, -s can be included to provide that discrimination or --from/--thru to
select a subset of the timeframe. Note that if one specifies
subsystems for which data has not been recorded, they will be displayed as zeros.
One can also change the format that the data is displayed though
various switches such as --verbose and -o.
-p file -P
Data is played back from the raw file and displayed on the
terminal in Plot Format. Note that since one often uses this mode to produce
output usable by other tools/programs, the user can force the output format by
including -s and only those subsystems specified will be displayed.
Furthermore, subsystems for which data has not been collected will also be displayed
as zeros to ensure consistent formatting across multiple data files.
-p file1 -f file2
This is NOT supported as you can only write data that is played
back to another file in plot format. Someone wanting to do this should rethink what
it is they are trying to do.
-p file1 -f file2 -P
Data is played back from the raw file and written to
the specified file in Plot Format. Note that here too -s will force specific subsystems
to be displayed.
Data is reported in this form when either --verbose is used OR if there is at least one
type of data requested that doesn't have a brief form such as any detail data or
ionodes, processes or slabs. Specifying some of the lustre output options with --lustopts
such as B, D and M will also force verbose format.
This table shows the total number of memory fragments by pagesize in increasing powers of 2 for
all the memory types.
CPU, collectl -sc
# CPU SUMMARY (INTR, CTXSW & PROC /sec)
# USER NICE SYS WAIT IRQ SOFT STEAL IDLE INTR CTXSW PROC RUNQ RUN AVG1 AVG5 AVG15 RunT BlkT
These are the percentage of time the system in running is one of the modes, noting that
these are averaged across all CPUs. While User and Sys modes are self-eplanitory, the others
may not be:
User
Time spent in User mode, not including time spend in "nice" mode.
Nice
Time spent in Nice mode, that is lower priority as adjusted by
the nice command and have the "N" status flag set when examined with "ps".
Sys
This is time spent in "pure" system time.
Wait
Also known as "iowait", this is the time the CPU was idle during an
outstanding disk I/O request. This is not considered to be part of the total or
system times reported in brief mode.
Irq
Time spent processing interrupts and also considered to be part of
the summary system time reported in "brief" mode.
Soft
Time spent processing soft interrupts and also considered to be part
of the summary system time reported in "brief" mode.
Steal
Time spend in involuntary wait state while the hypervisor was servicing
another virtual processor.
This next set of fields apply to processes
Proc
Process creations/sec.
Runq
Number of processes in the run queue.
Run
Number of processes in the run state.
Avg1, Avg5, Avg15
Load average over the last 1,5 and 15 minutes.
RunT
Total number of process in the run state, not counting collectl itself
BlkT
Total number of process blocked, waiting on I/O
Disks, collectl -sd
If you specify filtering with --dskfilt, the disks that match the pattern(s)
will either be included or excluded from the the summary data. However, the data will
still be collected so if recorded to a file can later be viewed.
Read requests merged per second when being dequeued.
Reads
Number of reads/sec
SizeKB
Average read size in KB
KBWrite
KB written/sec
WMerged
Write requests merged per second when being dequeued.
Writes
Number of writes/sec
SizeKB
Average write size in KB
Inodes/Filesystem, collectl -si
# INODE SUMMARY
# Dentries File Handles Inodes
# Number Unused Alloc % Max Number
40585 39442 576 0.17 38348
DCache
Dentries Number
Number of entries in directory cache
Dentried Unused
Number of unused entries in directory cache
Handles Alloc
Number of allocated file handles
handles % Max
Percentage of maximum available file handles
Inodes Number
Number of inodes in use
NOTE - as of this writing I'm baffled by the dentry unused field. No matter how
many files and/or directories I create, this number goes up! Sholdn't it go down?
Count of current errors. Since these
are typically infrequent, it is felt that reporting them as a rate would result
in either not seeing them OR round-off hiding their values.
Lustre
Lustre Client, collectl -sl
There are several formats here controlled by the --lustopts switch. There is
also detail data for these available as well. Specifying -sL results in
data broken out by the file system and --lustopts O further breaks it out by OST.
Also note the average read/write sizes are only reported when --lustopts is not specified.
This display shows the size of rpc buffer distribution buckets in K-pages. You can find the
page size for you system in the header (collectl --showheader).
RdK
KBs read/sec
Rds
Reads/sec
nK
Number of pages of of this size read
WrtK
KBs written/sec
Wrts
Writes/sec
nK
Number of pages of of this size written
Lustre Meta-Data Server, collectl -sl
As of Lustre 1.6.5, the data reported for the MDS had changed, breaking out the Reint
data into 5 individual buckets which are the last 5 fields described below. For earlier
versions those 5 fields will be replaced by a single one named Reint.
Number of getattr calls, for example lfs osts.
Note that this counter is not incremented as the result of ls - see Gxattr
GttrLck
These are getattrs that also return a lock on the file
StatFS
Number of stat calls, for example df or lfs df.
Note that lustre caches data for up to a second so many calls within a second may only show
up as a single statfs
Sync
Number of sync calls
Gxattr
Extended attribute get operations, for example getfattr,
getfacl or even ls. Note that the MDS must have been mounted with -o acl
for this counter to be enabled.
Sxattr
Extended attribute set operations, for example setfattr or setfacl
Connect
Client mount operations
Disconn
Client umount operations
Create
Count of mknod and mkdir operations, also used by NFS servers internally when creating files
Link
Hard and symbolic links, for example ln
Setattr
All operations that modify inode attributes including chmod, chown, touch, etc
Rename
File and directory renames, for example mv
Unlink
File/directory removals, for example rm or rmdir
The following display is very similar the the RPC buffers in that the sizes of different size
I/O requests are reported. In this case there are requests sent to the disk driver.
Note that this report is only available for HP's SFS.
# MEMORY SUMMARY
#<-------------------------------Physical Memory-------------------------------------><-----------Swap------------><-------Paging------>
# Total Used Free Buff Cached Slab Mapped Anon Commit Locked Inact Total Used Free In Out Fault MajFt In Out
Total
Total physical memory
Used
Used physical memory. This does not include memory used by the kernel itself.
Free
Unallocated memory
Buff
Memory used for system buffers
Cached
Memory used for caching data beween the kernel and disk, noting direct I/O does not use the cache
Slab
Memory used for slabs, see collectl -sY
Mapped
Memory mapped by processes
Anon
Anonymous memory. NOTE - this is included with mapped memory in brief format
Commit
Accorting to RedHat: "An estimate of how much RAM you would need to make a 99.99% guarantee
that there never is OOM (out of memory) for this workload."
Locked
Locked Memory
Inactive
Inactive pages. On ealier kernels this number is the sum of the clean, dirty
and laundry pages.
Swap Total
Total Swap
Swap Used
Used Swap
Swap Free
Free Swap
Swap In
Pages swapped in/sec
Swap Out
Pages swapped out/sec
Fault
Page faults/sec resolved by not going to disk
MajFt
These page faults are resolved by going to disk
Paging In
Total number of pages read by block devices
Paging Out
Total number of pages written by block devices
Notes
If you include --memopts R, memory and swap values wil be displayed as changes/sec between intervals
rather than absolute values in addition to page fault information, which is already displayed as rates.
This switch will also honor -on in that the values will not be normalized to a rate but rather displayed
as changes in size per interval.
If you include --memopts with P or V, collectl will only display Physical or Virtual memory.
The default is PV and will display both.
Memory, collectl -sm --memopts ps
The p and s options allow you to display data about page and/or steal and scan information. If you want this data combined with
the standard physical or virtual data you must explicitly request them as well. The columns show how the memory is allocated
for the respective sections.
The entries for error counts in the following are actually the total of several
types of errors. To get individual error counts, you must either include --netopts e
or report details on individual interfaces in plot format by specifying -P.
Transmission errors are categorized by errors, dropped, fifo, collisions and carrier.
Receive errors are broken out for errors, dropped, fifo and framing errors.
If you specify filtering with --netfilt, the names that match the pattern(s)
will either be included or excluded from the the summary data. However, the data will
still be collected so if recorded to a file can later be viewed.
Total incoming errors/sec. This is an aggregation of incoming error counters. To see explicit
error counters use --netopts e
KBOut
Outgoing KB/sec
PktOut
Outgoing packets/sec
SizeO
Average outgoing packet size in bytes
CmpO
Outgoing compressed packets/sec
ErrsO
Total outgoing errors/sec. This is an aggregation of outgoing error counters. To see explicit
error counters use --netopts e
Network, collectl -sn --netopts e
This alternative format, which is displayed when you specify --netopts e enumerates
the individual error types. You cannot see both output formats at the same time.
As of version 3.2.1, by default collectl collects and reports on all versions of nfs data,
both clients and servers. One can limit the types of data reported with --nfsfilt
and if only server or client data has been selected, only that type of data will be
reported as shown in the 2 forms below. When both server and client data are being
reported they will be displayed side by side. As with brief format, if filters have been
selected they will be displayed in the header.
# NFS SUMMARY (/sec)
#<---------------------------server--------------------------->
# Reads Writes Meta Comm UDP TCP TCPConn BadAuth BadClnt
Reads
Total reads/sec
Writes
Total writes/sec
Meta
Total nfs meta data calls/sec, where meta data is
considered to be any of: lookup, access, getattr, setattr, readdir and readdirplus,
noting that not all types of nfs version report all as V3 clients/servers do.
Comm
Total commits/sec
UDP
Number of UDP packets/sec
TCP
Number of TCP packets/sec
TCPConn
Number of TCP connections/sec
BadAuth
Number of authentication failures/sec
BadClnt
Number of unknown clients/sec
# NFS SUMMARY (/sec)
#<----------------client---------------->
# Reads Writes Meta Comm Retrans Authref
Reads
Total reads/sec
Writes
Total writes/sec
Meta
Total nfs meta data calls/sec, where meta data is
considered to be any of: lookup, access, getattr, setattr, readdir and readdirplus,
noting that not all types of nfs version report all as V3 clients/servers do.
Comm
Total commits/sec
Retrans
Number of retransmissions/sec
Authref
Number of authrefreshes/sec
NFS, collectl -sf -nfsopts C
The data reported for clients is slightly different, specifically the
retrans and authref fields.
As of the 2.6.22 kernel, there is a new slab allocator, called SLUB, and since
there is not a 1:1 mapping between what it reports and the older slab allocator,
the format of this listing will depend on which allocator is being used. The following
format is for the older allocator.
# SLAB SUMMARY
#<------------Objects------------><--------Slab Allocation-------><--Caches--->
# InUse Bytes Alloc Bytes InUse Bytes Total Bytes InUse Total
Objects
InUse
Total number of objects that are currently in use.
Bytes
Total size of all the objects in use.
Alloc
Total number of objects that have been allocated but not necessarily in use.
Bytes
Total size of all the allocated objects whether in use or not.
Slab Allocation
InUse
Number of slabs that have at least one active object in them.
Bytes
Total size of all the slabs.
Total
Total number of slabs that have been allocated whether in use or not.
Bytes
Total size of all the slabs that have been allocted whether in use or not.
Caches
InUse
Not all caches are actully in use. This included only those with non-zero
counts.
Total
This is the count of all caches, whether currently in use or not.
This is format for the new slub allocator
# SLAB SUMMARY
#<---Objects---><-Slabs-><-----memory----->
# In Use Avail Number Used Total
One should note that this report summarizes those slabs being monitored. In general
this represents all slabs, but if filering is being used these numbers will only
apply to those slabs that have matched the filter.
Objects
InUse
The total number of objects that have been allocated to processes.
Avail
The total number of objects that are available in the currently allocated slabs.
This includes those that have already been allocated toprocesses.
Slabs
Number
This is the number of individual slabs that have been allocated and
taking physical memory.
Memory
Used
Used memory corresponds to those objects that have been allocated to
processes.
Total
Total physical memory allocated to processes. When there is no filtering
in effect, this number will be equal to the Slabs field reported by -sm.
Sockets, collectl -ss
# SOCKET STATISTICS
# <-------------Tcp-------------> Udp Raw <---Frag-->
#Used Inuse Orphan Tw Alloc Mem Inuse Inuse Inuse Mem
Used
Total number if socket allocated which can include additional types such as domain.
Tcp
Inuse
Number of TCP connections in use
Orphan
Number of TCP orphaned connections
Tw
Number of connections in TIME_WAIT
Alloc
TCP sockets allocated
Mem
Udp
Inuse
Number of UCP connections in use
Raw
Inuse
Number of RAW connections in use
Frag
Inuse
Mem
TCP, collectl -st
These are the counters one sees when running the command netstat -s, whose output is very verbose. Since this
format is an attemt to compress those field names to 6 characters or less, sometime something gets lost in the translation.
As described in the brief data formats, the actual TCP data displayed is based on the value of
--tcpfilt and like brief data, everything is displayed on a single line which can be quite wide, even more reason to
use this switch, espcially since the default format is over 200 columns wide! The following definitions are based the
value of that filter:
One of collectl's main features is its ability to generate files in a ready-to-plot format which
is compatible with what gnuplot expects and there are actually 2 main types of files that it
generates. The first, which has an extension of tab, represents a table of all the summary
data. What makes this file unique is that all data elements are in a fixed set of columns -
some columns may get added over time, but for all intents and purposes, the set of data for say CPUs
do not change regardless of how many CPUs are in the system. The second type of files deal with
detail data, the amount of which changes with the number of instances so a 4 CPU system will have 1/2
the data an 8 CPU system has. There is one file for each type of detail data.
Plot files can be generated in 2 ways and each has its own advantages as well as disadvantages.
Post processing - collectl runs with minimal overhead, writing all its data to raw files.
Those files are then converted to plot files at some time in the future
During collectl - in this mode, collectl writes directly to plot files eliminating the need
for conversion of raw file
At first glance, it sounds like you'd always want to generate plot files directly since you avoid
the need for the conversion step, but you should also realize a few things about this methodology:
When plot files are generated directly you no longer have access to the original data. This means
you can't play back the data over selected periods of time nor can you select different data to examine,
for example if you chose to record CPU summary data but later decide you want to see CPU detail data.
In some cases you may want to look at data in unnormalized form and cannot
Some data is never converted to plot format and therefore is lost forever
With raw data you can always see the data in its original format if any questions arise to its accuracy
You can play back data multiple times, generating different views as well as plot files as often as you like
You can select different time intervals over which to play back the data
Playing back raw data allows you to display it in multiple formats such as --export vmstat or use
--top to look at process data in different ways
Most important, if you really want the best of both worlds you can record data in plot format and
with the use of --rawtoo also record the data in raw form.
Generating Plot Files On-The-Fly
While generating files this way is as easy as appending -P to the collectl command either
when run interactively or in /etc/collectl.conf, there are a couple of things to keep in mind:
If you want immediate access to the data while collectl is running be sure to always flush
the buffers and don't compress the data by including the switches: -F0 -oz
Compressed data takes about 90% less storage, so this may be an option too
Be sure to explicitly list all the subsystems you want plots for. In other words if you want CPU
detail data, be sure to include C with the subsystem selection. If you want both summary and
detail CPU data you'll need cC.
If you're afraid you'll lose critical data, consider using --rawtoo
Generating Plot Files from RAW Files
Collectl has the capability to play back a single file or multiple once
but in either case
the first thing collectl does is examine the raw file header to get the
source host name and creation date. There will always be a new set of data
generated for each unique combination of host and creation date. Note that
depending on the subsystems chosen there may be multiple output files generated.
This also means a single raw file that spans multiple
dates will result in a single set of data.
By default, the name of the plot file contains only the date and a test is made
to see if a file with that name already exists. If not, it is created in
append mode. This means that multiple raw data files for the same
host on the same date will result in a single set of data. However, if that
file already exists, collectl will NOT process any data, and request you
specify -oc to tell it to perform the first open in create mode
so that subsequent files can be appended. If you specify -oa
all files will be appended to the original one which may not be what you want.
Collectl cannot read your mind so to be safe, be explicit. If you want to
generate a unique set of data files for each raw file use -ou
which causes the time to be included in file names, resulting in a unique output
file name for each raw file.
This certainly maximizes your flexibility for all the reasons listed earlier. However, this now puts
the responsibility of managing your data more squarely on your shoulders. Some of the questions you need
to answer include:
Do you want to convert the raw files to plot files every day or just when needed?
Where do you want to store the plot files and how will you get them there?
Will you automate the file copies/conversion via a cron job or do it manually when needed?
Should you always convert everything to plot files or just do summary data, only generating detail
data when needed?
As with on-the-fly generation, should the plot files be compressed or not?
Having answered these questions and perhaps others, it now just becomes a matter of executing
the appropriate copy and/or collectl commands, which can be relatively easily scripted.
TIP - If you rsync raw files to another server and then process
them using a wildcard in your playback command, you will probably end up processing some of today's files too!
If you then later copy over the rest of today's file(s) you will need to recreate today's plot file since
collectl will not overwrite an exiting file by default. But if you specify the -oc switch with a wild
card you will end up recreating all the plot files which will result in a lot more processing
than you were planning on. Collectl supports a special syntax that allows you to playback just the
files from yesterday by replacing that string with yesterday's date as in the following:
collectl -p "YESTERDAY*" etc...
noting that all uppercase characters are required and you can include other characters in the string
such as a host name if need be.
TIP - If you want to create multiple sets of plot files from the same raw file, you can always
include a unique qualifier along with the directory name with the -f switch to give each set a different
prefix.
Daily vs Unique Plot Files
Collectl raw files are created every time a new instance of collectl is run or whenever
collectl is instructed to create new one via -r such as when running as a Daemon. This
is why each file name include a time as well as a date. However
trying to plot multiple files for any given day can be problematic even for an automation script
that might help generate plots for you and so by default collectl creates non-timestamped, daily
plot files.
Whether you choose to create plot files on-the-fly or manually (by playing back existing files),
if you've not instructed collectl to do anything with unique file, it will simply append new data
onto an existing file. One the other hand if you explicitly ask for unique files, whenever
a new raw is processed or a new instance of collect is run, a new plot file will be created
that includes a corresponding timestamp.
The obvious question then becomes, why would you ever choose to create unique files when they're
such a pain to plot? and there are actually several good reasons you might choose to do so:
If periodically rsycn plot files to another system, one large file is always changed
and must be copied in full. When there are multiple files only the latest one is ever copied.
If generating a single plot file from a raw file multiple times/day to be more dynamic,
collectl always has to process the full raw file, which will take longer as the file grows. If you
create multiple raw files only the latest will need to be reprocessed.
But the main reason, and one that requires more explanation, is dealing with
configuration changes
Dealing with configuration changes Consider the following situation: you run collectl twice during the same day with the following commands:
collectl -scd -P -f/tmp
collectl -scdm -P -f/tmp
or perhaps you even generated raw files first and later play them back, converting them to
plottable files. In either case you now want to plot the data. Since it's all in the same file, the
headers that are initially written will only tell you there is cpu and disk data in the file! Collectl
will have written a second set of headers in the file at the time the second instance was run, but do
you really want to have to make a pass through the whole file every time you want to generate a plot
looking for additional data? Furthermore, there will be less columns of data in the first part of the
file that the latter, a condition that will probably cause most plotting packages to blow up, so you
really need unique files.
Another situation that can cause this is when dealing with detail data for dynamic subsystems such as
Lustre and disks. Dynamic change detection for Lustre has always been a part collectl but
support for dynamic disk configuration changes has been added to collectl V3.3.4. When a
configuration change is detected, it forces collectl to create a new file, whether generating data
in raw or plot format. Furthermore, if you later try to combine disk data from multiple raw
files into a single plot file that has disk configuration change data in it, collectl won't let you
unless you specific -ou forcing the generation of unique files. Lustre changes can be combined into
a single plot file by adding in any missing columns and 0-filling them.
Caution
If you are generating non-unique plot files on-the-fly and a
configuration change occurs, that new data will simply be appended to the single file.
There is nothing collectl can do about this, because it wants to keep all files
in a consistent name format and will not switch to unique name formatting without
explicitly being told to do so.
Configuration changes do not happen often and this is only an issue when generating
plot files in real-time. Furthermore, since this only effects detail, because
the summary data will always accurately reflect the sum of the instance data, this typically
will not effect anyone but is being stated for completeness.
If after all this you choose to generate real-time, non-unique detailed plot files and find
yourself in a situation where you should have, you can always write a script to split the
plot files back into individual ones since there is sufficient data in the internal headers
to do so. If you have multiple unique files and find single files easier to deal with, you
can also choose to write a post-processing script that merges these into a single file with
zero-filled columns where there is missing data.
updated June 4, 2009
collectl-3.6.9/docs/Ganglia2.jpg 0000644 0001750 0001750 00000023527 12230241726 014564 0 ustar mjs mjs JFIF ` ` C
$.' ",# (7),01444 '9=82<.342 C
2! !22222222222222222222222222222222222222222222222222 9"
} !1AQa"q2#BR$3br
%&'()*456789:CDEFGHIJSTUVWXYZcdefghijstuvwxyz
w !1AQaq"2B #3Rbr
$4%&'()*56789:CDEFGHIJSTUVWXYZcdefghijstuvwxyz ? (
(
(
(
(
(
+7Z=WI e!v 8㟡;>uEw 4o\ y_K?Y5懨y
6
aqE_x{C&f'bVvlu!Tr=hzn ]+%xGT.5+Ma^XVT ;pO~*[vRWNɐ7#PcEcxsz ldruX?* Hyuޞf6xQX Ef^|2 (i VG+)SU# Cb Qυ5?B*. c6D& c~9A ^W(%M$Ɠv7ATWA|QM\Nɣ "6= 8=*=S: i̖b5C(cpdUEuCh7m:-CkPKim!m}]ޭ>]kgYOjZZ+~(7ZcL?L*@:
l_|6kb!Z Q 3ڧو4\։¾#ҴfCU` V +I㞔|Ef+{PP*MqGo0uW㏉g·ں7~sTdzӽ;E1B H$ ]d1:_ m:+?E4iqjZMwVrl8 Т ( ( ( ( ( ( ( (:i::nDR\cp,@a6]/Lxn}2H-oŬv]8z->} WDarecP|"&x
Eӵ+k2oƉdaX\2BT<½=_zϊ״n->u DN|}rrIځS ˄Tvc@үbTվ5G%F28#s]4:OWe+-Lc mڵt ^kǹ>dBc cz S)9?7zY;>x7
͌b?,i迫[-R}ϻ}_{I"+ U hd>?|7?µ-t"k1h4oH-w$2^xx~|cz ~N # W <^h
p1 7vZiÕNSW}^FMSPj1sqi1gPؕ|gsCoe5ύ~$az}skO eY.
W&?/ nlZ4FժHYC+L z(ia&QN;Mt[䚿<9wXZOͅ%ő|ȌF<9|u&kǚ Lj UPnu!gܨxx~|cz ~N #-`5E1&I[s1'$O@ T{7̗Bo>M:8/c DoS9>]'OB3bkz%ޟ"-^]AeV+{-Z7wks+O ? |w_mӮU1`Pm#8qcKؼ%cu(NQ4 eI Ȫ 23+Ы&5vڀNT\8spVY켟r嫲k8 x^k=ֱ%[M6 A
Upq q/`g τt2cH$fsj:ψ>4QQy6ko.$(+HOA 2ld^7fm7F:ۍyg5]i{ dy~^v.& )`|Oiȼoo&nt ד [6 m&@q]y;~O.o'ӿ^Οo0mß
u+cjWOjb8w?劁F2zK_ {kh:=C]I+o+.TpsWEn6tc]͍'d^7fm7F:ۍymMoXJ5o_nWM -qq-ćZ}Fw9='"7Mh 1.lW^Nߓ֪Klg8F;J(
(
(
(
(
(
(
(
(
(
(
/MBkJ݀5TuZB[<[X7|Ԟz w=5oKMz h Tc?<>#kgn+> + 3X
<=sKsǚVImla eϷzOVDKƻcٸrq ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( +~%^\xwÝ.V[X\';r }+мwxJXkL˶ IO N\ Ox~m{Viy̨NUO9,};PYY0YZD[ƱU TQ@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@W]ifv-Gˌ3~@ =( qmG kT$3x ~G?m lT( 3wWxG A~_wWxG A~_wWxG A~_wWxG A~_wWxG A~_wWxG A~_wWxG A~_wWxG A~_wWxG A~_wWxG A~_wWxV<;ki 4OUrdy4ri \Ey?b
ÿ 1W $nQEQ@Q@Q@Q@yOOhV_Wȍʄlhڻ}:{۩vFde&NjUN2#,`boA#8 ҡ:. C`+r( ( ( ( ( ( ( ( ( ( ( +Js ZUNo *Йt;(=}x7?WWw>C a*_č(( ( ( G-/۽ZBL#G4|J.|Ki d"KYu q@?͏}kt>Im8v_akUnUmerz/z- QE QE QE QE QE QE QE QE QE QE QE No +Jt) `c \:.qET f* 3s\~KETT4:k(d?+ O ~ ~W ~8 > Ů&<.fU0RN:O ?> ߕ ⫔ hgTui T} ?S + W)E 4{ X *:O kƾ5oAӈ#3 lkjO(ѱ#ǚj(Uu%P0 @ O ~ ~W R4{u _'?U i UQGM ?: ʎ?_*i1cYCl3F U G
'X2TJiwW)E$K+>+Ɲky=/h$1E+ |W !m[ Ыz2<,% i}N?_*O (W&v| _Q } ?S + Q @O \dc > ߕ O ~ ~W R?haTui T} ?S + W)E 4{ X *:O ?> ߕ ⫔=u _'?U i UQGM ?: ʎ@Qs4+ WY^= >_1XazU25k;f_S G
Ҭ+Js \pzRwQEAA^{nB=}x7?UdQEAG? Zk F-q5 O_
(6wui_ެt2 ?GϧZ f| QS௱]c7bv fYI qJmiMH]ۉP 8'ְi>^ݺ#6Dkg&뜅j[ꖚ-w3A(M^j4
g.c&s7ozΚG]oЅWi-,י]N6blu&.Q!G Q$w`-4[ĵK D 8`ZX\ֶWԟ'72oIr6 jj^o67ݎ U vHɜ$f 8|aiPJcPL-<}k>$ | :(ՕI+'1hQ{uoB I ,JKyole-bU yj< =ZXy =JK^IX*$z.4I L 9~u?k #K_# a#y9 gz'WZi~me'93}{ ^S7o B) Be 핓7nU[[ּ
9u=^R |?>hѢ+ψ
+Fo.o$pZ7*`^-N׆n5tQerNsClp9wTn`=N+bmʿm~ӟDd }Τ-KMY,&Dvd;ch{?+9דM mO ~mu5'/(Jd!4s(ORZ] h[ǹid[i/f4Mf{O2gSjN7ޱ6;@^:Gn;Yt$utw, [Kﷶ/m춏by#ךRRZ/t c.kݵlNb(*jb{S |5g ď\9?_f_S G
Ҭ+Js ^T: QPPWw>C a+Ы|; ^! qY/FQPQD g q "} ы\M{?/SqiXk\^1b7A@,^"-x Z
ZZ _yM^k\]m#r\n
H
@$''(_̓P_>H6+ 9#h ~hcVoiNN=z˪]RW}5$_N&s F<4,F
pOsyz<"6 hi8ǘH#k Ow7QJX.oo]wipͣxQ۬miN=J+ _3G. w肳t) `c iVn %9 p* }~]((+|; ^! U f,#r((il7 PF# A
O / ]]*
v3u5 CS A
O / ]]Y?}) A
O / G :5?MutQ?}) A
O / G :5?MutQ?}) A
O / G :5?MutQ?}) A
O / G :5?MutQ?}) A
O / VtX隔7^,;V^AGc]zVreGJ.L(+6sy=/iiYcB'' -tU¤b'N3V CS A
O / ]]k32_q CS A
O / ]]}f0_q CS A
O / ]]}f0_q CS A
O / ]]}f0_q CS A
O / ]]}f0_q CS A
O / ]]}f0_qɷ5kjLi YEN)t) `c iVn %9 p }.qET f* 3s\~KETQE QE QE QE QE QE QE QE QE QE QE QE QE QE QE QE No +Jt) `c \:.qET f*l|MjP9~ yVNFS Monitoring
NFS Monitoring
Introduction
As of version 3.2.1, nfs monitoring has undergone a major change. Unlike the old behavior which
only reported on the client or server for a specific version, by default nfs reporting now collects
data for all types of nfs data including nfs version 4. This also means when reporting nfs
summary data, all statistics are aggregated in both brief and summary formats.
A system typically runs one version of NFS and usually
acts as a client or server and so by aggregating the data the numbers being reported in brief mode
will already be for a single type of data. Furthermore, in verbose mode the client/server data are
broken out so even if the system is acting as both a client and server you will be able to
differentiate the data being reported.
As an optimization as well as a convenience, collectl looks at the raw read/write fields for
each set of nfs data, which represent the totals since boot.
If those fields are both 0, it is assumed there is no nfs activity and the rest of
the statistics will also assumed to be zero. With the excepion of the detail format
descripted later, this all happens behind the scenes. However, just because those fields are
non-zero does not mean there is currently nfs activity. In fact if you mount a filesystem, write
to it and dismount it, those counters will remain non-zero until the system is rebooted. In other
words, this approach is not perfect but simply provided as a mechanism to help reduce the
need for user-specified filters to help focus the detail output.
Data Filtering
As they say, your mileage may vary and if your system is running mixed nfs versions and/or
acting as a client and server and you really only want a subset of the activity included in
the summary or detail reports, you can modify the behavior by specifying one or more filters with
the --nfsfilt switch. If you just select clients, only data for those clients will be included
in brief, summary and detail formats. However, if recording to a raw file, collectl
will record data for all NFS client versions even if you've only selected one or two. The same holds
true for servers. When you do use -nfsfilt, those values are display in the brief and verbose
headers both during collection and playback.
Detail Data
The data collected for NFS V2 and V3 is quiet similar in that V3 reports of superset of what V2 reports
with the exception of root and wrcache. V4 reports a lot more counters than either
but many of the same key ones as V3 and so the detail format has been standardized on the V3 counters.
In this mode, one line is reported for each of the 6 types, with blank entries for the non-common fields
as shown here. Note that only rows for those types determined to be active by checking the
read/write fields or explicitly selected through filters, will be included:
Looking at detail data for more than one type of data can be difficult to watch.
Consider using --home which can give the feel of a real-time display in top format.
Playing back data generated by older versions of collectl
Collectl is smart enough to do the right thing. In other words if you're playing back data generated
by a pre-3.2.1 version of collectl, collectl figures out what type of data the file contains and actually
sets --nfsfilt for you (in fact it won't let you select it yourself) and only displays the type
of data in the file.
Playing back newer data with older versions of collectl
Any raw file created by Version 3.2.1 or greater records nfs data in a format that will not
be recognized by earlier versions of collectl and any attempts to read it will result in fields of all
zeros. It is not expected that this would typically happen but as they say, it is being stated here
for completeness.
The following is a list of the files read by collectl to support the different types of data
being collected. Also included are the basic linux commands that
should produce the same numbers as collectl for those times you may want to know if
you've uncovered a collectl problem OR it's a linux problem. The one exception is
Infiniband data which is obtained by the perfquery OFED utility as noted.
When something that may be of interest occurs, collectl calls an internal message reporting
routine and assigns that message a status of Informational, Warning, Error or Fatal.
In all cases, if a message of type Fatal is encountered, collectl will terminate.
In all other cases it continues executing, often skipping what it was trying to do
when the error occurred.
The way collectl deals with these messages is controlled by several factors:
Interactive Messages
Informational messages are never displayed on the terminal.
Warnings are only displayed if --quiet has not been specified
Errors and Fatal messages are always reported
If -m is specified, ALL messages are displayed on the terminal. When collectl is not
providing the desired results and it is not obvious why, adding this switch can be helpful.
Daemon Mode Messages, requires -f and -m
If collectl is started as a daemon and the -m switch is specified, all messages will be
written to the message log in the collectl logging directory, the default being /var/log/collectl.
If this switch is not specified no messages are ever recorded and so it is recommended that
this switch, which is already in the collectl startup script, not be removed.
Any messages of type Error or Fatal are also sent to syslog, but again only
if logging is enabled with -m.
updated Feb 21, 2011
collectl-3.6.9/docs/Disks.html 0000664 0001750 0001750 00000012270 12230241727 014375 0 ustar mjs mjs
collectl - Disk Info
Disk Monitoring
Introduction
As with other subsystems that have specific devices, collectl can report disk summary as well as detail data.
Like other summary data, disk summary data represents the total activity across all disks or to be more precise, those
enumerated in /proc/diskstats, with a couple of caveats:
Partition data is skipped
Only those disk names that explicitly match the pattern in DiskFilter (see collectl.conf) are included. But don't
worry, all major types of disks are included as well as several others as they are identified.
Device mapper disks, while listed in the detail data are NOT included in the summary data to avoid
double counting
The three key counters for disk activity are bytes, iops and merges, though only bytes and iops are reported
in brief summary mode. The average I/O size which is also reportd in verbose and detail modes may be optionally
included in brief mode by including --iosize or --dskopts i. If you're not sure why/when you'd care about summary
data, be sure to read this.
Disk detail goes a step further and in addition to including the same information that's reported for summary data
also includes key device specific metrics relating to queue lengths, wait and service times as well as utilization.
For those familiar with iostat, this is the same data it reports. These numbers will help you determine if your
individual disks are operating properly since high wait and/or service times are a bad thing and indicate something
is causing an undesired delay somewhere.
Filtering If you'd like to limit the disks included in either the detail output or the summary
totals, you can explicity include or exclude them using --dskfilt. The target of
this switch is actually one or more perl expressions, but if you don't know perl all you
really need to know is these are strings that are compared to each disk name. If the first
(or only) name is preceded with a ^, disks that match the string(s) will be excluded.
Dynamic Disk Discovery
Dynamic disks are handled by the exact same algorithms that are applied to dynamic networks and
while they have not yet been found to have the same problems as netoworks do with potentially
hundreds of orphaned names no longer in use, the same logic for dealing with stale disks has
been added to netowrk processing data and rather than be repetitious, read the descrption
for dynamic network processing and learn how the new disk
option --dskopts o would be applied.
Included with collectl is the example file hello.ph which is a collectlized version of
hello world. It simulates a hw subsystem consisting of 3 hw instances, which in turn
report a single counter. Here is an example of its simulated /proc data, which is shown by
using -d4. Also notice in this case the discriminator is hw but -n is also
included in the calls to record() to further identify individual devices:
You can use this example module with virtually any combinations of switches and any other collectl
subsystems as well as exporting the output over a socket, writing to a raw file or playing it back.
As you should realize by now the combinations are far too extensive to list so below is only the
simplest one, showing this data combined with cpu stats in brief format with timestamps in msecs.
collectl --imp hello -sc -oTm
# <--------CPU--------><-Hello->
#Time cpu sys inter ctxsw Total
11:40:29.002 0 0 1027 126 140
11:40:30.002 0 0 1012 138 230
For further information on using this capability see hello.ph which has been heavily annotated and
should make a good staring template for developing your own custom modules.
There has been a recent problem identified when running collectl on systems
with versions of the Time:HiRes perl module less than 1.91 and versions
of glibc newer then 2.3. So far these messages appear to be harmless as
they've only been identified as occurring during system boot. However, since the
problem is that HiRes is actually calling the setitimer system
service incorrectly, it is really living on borrowed time (if you'll pardon the
pun) and users would be much better off and safer to simply upgrade to a newer
version which they can get at http://search.cpan.org/dist/Time-HiRes/.
If you've never installed something from CPAN before you shouldn't be intimidated
even if you're not a programmer as all you need to do is perform the following
steps. If the version you're downloading is not 1.9715, replace that string
in the instructions below with the appropriate version number:
download Time-HiRes-1.9715.tar.gz to /tmp
cd /tmp
unpack it with the command tar -zxf Time-HiRes-1.9715.tar.gz
cd Time-HiRes-1.9715
perl Makefile.PL
make
make test
make install
collectl -v should identify the newer version
In a few cases, rather than replacing the older version the new one ends up in a
different location, and collectl still sees the old version. You can usually get
around this problem by re-executing the make install command as make
install -UNINST=1
Update - April 08, 2009 It looks like RedHat has finally responded to my bugzilla and posted a
solution, which makes
me optimistic we should see a newer version of Time::HiRes in the RHEL5.4 timeframe.
That still doesn't mean the problem has been resolved on distros that use older versions.
As with other subsystems which contain instance data, you can monitor both summary
(in brief and verbose modes) and detail data. Like disk data, the key brief mode values are bytes and packets
(rather than iops). The actual data comes from /proc/net/dev.
The one key thing to keep in mind with network data is that not all networks are the same. Just like there are
device mapper disks that shouldn't be included in the summary data the same is true for network devices. Those that
are not included in the summary are the loopback, sit, bond and vmnet devices,
Since most lan networks run fairly cleanly and errors are rare, one is usually not interested in seeing long columns
of zeros that never change and so by default brief mode does not include any error information. Adding
--netopts e will add an additional column with a total error count. To see specific errors one would have to
run in summary more and do identify the specific networks on which those errors were occurring you would have to
run in detail mode.
What about IB over IP?
Good question. When using Infiniband networking you typically get an IB network device created. So does this mean
IB traffice gets counted twice when you monitor both it and network data? As they say, it depends.
Some Infiniband data will indeed go over the native IB interface and never show up as network data. This includes
MPI traffic or lustre which uses the native IB transport. However, other uses of Infiniband may in fact be counted
as network traffic. BUT this is actually a good thing because if you're a heavy user of IB/IP and want to be able
to differentiate the native IB traffic from it, simply look at the network detail data and subtract any IB network
numbers from the native values.
Tips and Tricks
Ever try looking for a needle in a haystack, in this case maybe it's network errors? --network E works just
like its lowercase cousin except it tells collectl to only report intervals that have network errors in them. While
this can be extremely boring in real-time mode, consider what happens during playback. During the course of a day
you'll have 8640 samples but this switch will allow you to see the one that recorded the network error!
Filtering If you'd like to limit the networks included in either the detail output or the summary
totals, you can explicitly include or exclude them using --netfilt. The target of
this switch is actually one or more perl expressions, but if you don't know perl all you
really need to know is these are strings that are compared to each network name. If the first
(or only) name is preceded with a ^, names that match the string(s) will be excluded.
This switch works exactly the same way as --dskfilt so for use case examples
see Disk Monitoring.
Dynamic Network Discovery
Network devices typically don't change unless you install a new network card and reboot the machine,
in which case collectl when the list of known networks is created it's simply correct.
However, especially when running on a host that creates virtual machines, networks can come and go.
The way collectl does this is to simply maintain a dynamic list of the active networks and make sure
the statistics get associated with the correct network. As a way to preserve network details in plot
format, collectl further retains a list of all the networks ever seen since system boot and uses that
list to record the statistics, thereby keeping the columnar data consistent. In many cases where
hypervisors reuse the virtual network names, the list of unique names remains relatively low.
However it was recently discovered that the OpenStack Quantum service for handling dynamic networks actually
uses a new network name every time a VM is created. If a particular host has dozens of VMs and they come
and go many times, the result can be a very long list of network names which not only show up in all the
headers of all the files collectl generates, they will also show up as unique sets of data in the network
detail file if one is generated. On one long running host, over 500 unique network names were generated!
The way collectl has chosen to deal with this is with --netopts o, which tells collectl to not
preserve the ordered list of all the networks that have been seen and will in fact cause collectl to prune
that list whenever a network is found to have been removed. While this solves the immediate need to keep
only the active networks in the headers of the output files it creates a different problem with the detail
files in that the data will no longer be consistent. In fact, it is probably best to not even try to
generate detail files for dynamic network when using this switch. As of this writing I'm still trying
to think of ways to improve the situation.
At the highest level, both collectl and sar provide lightweight collection of
device performance information. However, when used in a diagnostic mode sar
falls short on a number of points, though admittedly some could be addressed
by wrapping it with scripts that reformat the data:
sar plays back data by device/subsystem, sorted by time
sar does not deal with sub-second time
sar output cannot be directly fed into a plotting tool
sar does not provide nfs, lustre or interconnect statistics
sar does not provide for the collection of slab data
sar's process monitoring is limited in that if cannot save
process data in a file, cannot monitor threads, cannot select processes
to monitor other than ALL or by pid (so cannot selectively discover new
processes) and in interactive mode is limited to 256 processes
Isn't a default monitoring frequency of 1 second going to kill my system?
Running collectl interactively at a 1 second interval has been shown to provide
minimal load. However, for running collectl for long periods of time it is
recommended to use a default monitoring period of 10 second and in fact is the
default when collectl is run as a daemon and started using the 'service start
collectl command'.
A lot of effort has gone into making collectl very efficient in spite of the fact
that it's written in an interpretive language like perl, which by the way is known
for its efficiency. collectl has been measured to use less than 0.01% of the cpu
on most systems at an interval of 10 seconds.
To measure collectl's load on your own system you can use the command
"time collectl -i0 -c8640 -s??? -f." to see the load of collecting a day's worth
of data for the specific subsystems included with the -s switch.
What is the best monitoring frequency?
There really isn't a 'best' per se. In general collecting counter data every 10 seconds
and process/slab data every minute has been observed to produce a maximum amount of data
with a minimal load. When this granularity isn't sufficient there have been uses for
collecting data as 0.1 second intervals! There have even been times when wanting to
verify a short lived process really does start that doing process monitoring by name at
an interval of 0.01 seconds has been found to be useful.
Why so many switches?
In general, most people will not need most switches and that's the main reason
for 'basic' vs 'extended' help. However, it's also possible that there may be
an extended switch that provides some specific piece of functionality not there
with the basic ones and it is recommended that once you feel more comfortable
with the basic operations that you spend a little time looking at them too.
Why doesn't --top show as much data as the top command?
The simple answer is because this is collectl, not top.
Actually I thought of that and then decided with all the different switches
and options, the easiest thing to do is just run
a second instance of collectl in another window,
showing whatever else you want to see in whatever
format you like. You can even pick different monitoring intervals.
What does collectl stand for?
Collectl is based on the very popular collect tool written by Rob Urban which
was distributed as with DEC's Tru64 Unix Operating System and therefore stands for
collect for linux.
How do you pronounce collectl?
It rhymes with pterodactyl.
Why is the default socket port 2655??
Those are the first 4 digits of collectl on a telephone numeric key pad.
Running collectl
How do I get started?
The easiest way to get started is to just type 'collectl'. It will report
summary statistics on cpu, disk and network once a second.
If you want to change the subsystems being reported on use -s and to change
the interval use -i. More verbose information can be displayed with --verbose. See
the man pages for more detail.
How do I make a plot?
Collectl supports saving data in plot format - space separated fields - through
the use of the -P switch. The resultant output can then be easily plotted
using gnuplot, excel or any other packages that understand this format. You
can redirect collectl's output to a file OR it's much easier to just use the
-f switch to specify a location to write the data.
How do I drill down to get a closer look at what's going on?
The first order of business is to familiarize yourself with the types of data
collectl is capable of collecting. This is best done by looking at the data
produced by all the different settings for -s, both lower and upper case as
there is some detail data that is not visible at the summary level.
Take a look at -sd and -sD. If you still
don't see something it might actually be written in -P format. See -sT for an
example.
Next, run collectl and instruct it to log everything (or at least as much as you
think you'll need) to a file. When you believe you've collected enough data - and
this could span multiple days - identify times of interest or just plot everything
(see the -P switch).
Visually inspecting the plotted data can often show times of unusually heavy resource
loads. Often times there is a strong time delineation between good and bad.
It you want to see the actual numbers in the data as opposed to plots,
play back the data using the --from switch to
select a begin time, usually a few samples prior to the time when things started to
go bad. To reduce the amount of output you can also use --thru to set the end time for
the collection. You can also start selecting specific subsystems to look at as well
as individual devices. For example, if you've discovered that at 11:03 there was
an unusal network load, try 'collectl -p filename --from 11:02 --thru 11:05 -sN' to see
the activity at each NIC.
And don't forget process and/or slab activity if either has been collected. You can
also play back this data at specific time intervals too.
I want to look at detail data but forgot to specify it when I collected the
data. Now what?
Good news! With the exception of CPU data, collectl always collects
detail data whether you ask for it or not - that's how it generates the
summaries. When you extract data into plot format, by default it extracts the
data based on the switches you used when you collected it. So, if you
specified -sd you'll only see summary data when you extract it. BUT if you
include -s+D during the generation of plotting data you WILL generate disk
details as well.
How do I configure collectl to run all the time as a service?
Use the chkconfig to change collectl's setting to 'on'. On boot, collectl will
be automatically started. To start collectl immediately, type
'service collectl start'.
How do I change the monitoring parameters for the 'service'?
Edit /etc/collectl.conf and add any switches you like to the 'DaemonCommands'
line. To verify these are indeed compatible (some switches aren't), cut/paste
that line into a collectl run command to make sure they work before trying to
start the service.
What are the differences between --rawtoo, --lexpr and --sexpr
Looking at --lexpr and --sexpr first, these will cause the contents of most
counters to be written as a list or s-expression to either a file or socket
based on whether -f or -A is specified. If both are specified the data will
be sent over the socket and written locally as a raw file.
Adding -P will cause the local file to be written in plot format while adding
--rawtoo will cause both plot and raw files to be written locally.
How can I pass collectl data to other applications?
You actually have several choices, all of which are based on
--export in which you specify a routine that will export collectl's
output in some other format that your application
may prefer to see it in. There are currently 2 such routines:
lexpr.ph formats data as a list
sexpr.ph formats data as an s-expression
If you don't like either, you're free to write your own.
The next thing you need to decide is whether you simply want to write a
data snapshot to a local file which some other program/script can retrieve
OR send the data over a socket to your application. Clearly using the socket
is more efficient, but the choice is all yours.
An example parser script called readS had been provided as a convenient
way to parse a file that is an s-expression, but just keep in mind that it is
written in perl and every invocation involved starting up the perl interpreter
which may be a little heavier-weight than you wish to use.
If you do choose to use it, the arguments to readS take the following form:
dir category variable [instance [divisor]]
dir - directory in which the s-expression S exists, often
/var/log/collectl
category - category of data element, such asc cputotals or diskinfo
variable - the name of a summary item such as nice for cpu data
or the name of the detail category such as diskinfo.
instance - for detail data this is the instance name
divisor - divide the results by this number, noting for summary data
instance needs to be a null string
Detailed customization instructions for use of data returned by readS
is beyond the scope of this FAQ.
Gottchas
This section is for those things people might otherwise miss in the FAQ and Shouldn't, because
they can cause misunderstanding of what is being reported.
Round-off vs normalization
Since most of the numbers collectl reports are large, certainly greater than the number of
seconds in an interval, they are not affected and so this is rarely an issue. However there
are values that can be quite low such as the number of processes created/second or network errors.
This can also be the case for many of the nfs metadata operations counters, some of which report
values as low as 1 over many minutes.
The issue is if you're looking at data at some interval other than 1 second, the result of dividing
these numbers by the interval size can result in values less than 0.5 and since collectl rounds off,
those values would be reported as 0 and you could miss a critical data point. If you
tell collectl not to normalize the data, in other words NOT to divide by the interval, you'll
always see the actual numbers. However, since most people prefer values/sec, it's easy to
forget. Try not to...
Operational Problems
Why won't collectl run as a service?
As configured, collectl will write its date/time named log files to
/var/log/collectl, rolling them every day just after midnight and retaining
one week's worth. In addition it also maintains a 'message log' file named for
the host, year and month, eg hostname-200508.log - the creation of the message log is
driven off the -m switch in DaemonCommands. Check this log for any messages
that should explain what is going on.
Why is my 'raw' file so big?
By default, collectl will collect a lot of data - as much as 10 or more MB/day!
If the perl-Compress library is installed, these logs will automatically be
compressed and are typically less than 2MB/day.
The output file size is also affected by the number of devices being monitored.
In general, even on large systems the number network interfaces is small and
shouldn't matter, but if the number of disks gets very high, say in the dozens
or more, this can begin to have an effect on the file size. The other big
variable is the number of processes when collecting process data. As this
number grows to the many hundreds (or more), you will see the size of the data
file grow.
Finally the other parameter that effects size is the monitoring interval. The
aforementioned sizes are based on the defaults which are process/slab
monitoring once every 60 seconds and device monitoring once every 10 seconds.
Did you override these and make them too small?
Playing back multiple files to terminal doesn't show file names
By design, collectl is expected to be used in multiple ways and a lot of
flexibility in the output format has been provided. The most common
way of using playback mode is to play back a single file and therefore the name
of the file is not displayed. The -m switch will provide the file names as
they are processed.
Why don't the averages/totals produced in brief mode look correct?
There may be two reasons for this, the most obvious being that by default the
intermediate numbers are normalized into a /sec rate and the averages/totals
are based on the raw numbers. If the monitoring interval is 1 sec or you use
-on to suppress normalization, the results will be very close.
The other point to consider is that numbers are often stored at a higher
resolution than displayed and so there is less round-off error with the averages
and totals.
What does New slab created after logging started mean?
When collectl first starts, it builds a list of all the existing slabs.
As the message states, collectl has discovered a new slab and adds it to
its list. This is relatively rare but can also indicate collection was
started too soon, possibly before system processes or applications
have allocated system data structures. It is really just an informational
message and can safely be ignored.
Why does collectl say waiting for 60 second sample... but doesn't?
This is very rare as it will only happen when collecting a small number of process
or slab data samples, but it is also worth understanding what is happening
because it gets into the internal mechanics of data collection. In addition to the
normal counter collectl uses to collect most data, it also maintains a second
one for coarser samples such as process and slab data. When reporting how long
collectl is going to wait for a sample, it uses a number based on the type of
data being collected. In almost all cases this is the value of the fine-grained
counter, but if only collecting process or slab data, it reports the second
counter whose default is 60 seconds.
Collection of counters, such as disk traffic or cpu load, always requires 2
samples since it's their different that represent the actual value. Other
data such as memory in use or process data only require a single sample
but in order to synchronize all the values being reported, collectl
always uses its first sampling interval to collectl a base sample and doesn't
actually report anything until the second sample is taken which is why it
reports the waiting... message even if it isn't being asked to report
any counters.
Finally, the -c switch which specifies the number of samples to collect
applies to the finer-grained counter. This means if you try to collect a number
of samples that will cause the -c switch limit to be reached because any data
is actually collected, you will see collectl exit without reporting anything!
The best example of this would be the command collectl -sZ -c1.
Since the default interactive sample counters are 1 and 60 seconds respectively
and collectl has to actually take 2 samples, collectl will only run long
enough for one tick of the fine-grained counter or 1 second and immediately
exit with no output. Therefore to
collect 1 process sample you will actually need to use -c60 but will also have
to wait 60 seconds to see anything.
Alternatively you could set the fine-grained sample counter to the same as
the process sample counter
and so the command collectl -i60:60 -sZ -c1 would also report 1 sample
after waiting for 60 seconds. If you want to collect a sample after just 1
second, you should use collectl -i:1 -sZ -c1.
Why am I not seeing exceptions only with -ox?
Exception processing requires --verbose. Did you forget to include it?
I'm seeing a bogus data point!
This message means collectl has read a corrupted network statistics record
and is ignoring it. It also turns out this has been attributed to some
bnx2 chips and a workaround has been generated for newer drivers. If you
want a little more information your can read about it
here.
The way collectl determines a record is bogus is to look at the transmit and receive
rates for each interface and compare them to the speed of that interface
from /sys/devices OR if no entry found uses the value of DefNetSpeed
which can be overridden in collectl.conf).
It either exceeds twice the interface rate, the record is considered bogus
and ignored. This will cause collectl to report the previous rate
for this interval. While not foolproof, it is hoped this will reduce the
frequency of this type of data.
What does the error -sj or -sJ with -P also requires CPU details so add C or remove J. mean?
Interrupt reporting has a unique property in that summary data provides CPU specific data
while detail data provides data about individual interrupts and you will get this error
if you request interrupt plot data but not CPU detail data. The most common place this
can happen is if you run collectl V2.5.0 as a daemon because it collectl interrupt data
but not CPU detail data.
In order to play back plot data from a file that did not specify CPU details be
collected, you can either tell collectl not to include interrupts by the command
collectl -s-j... or tell it to also include CPU details with the command
collectl -s+C....
In order to make this less confusing with future releases, and until I think of simpler
way to do this, the collectl daemon will be set up to include CPU details, noting this
has no impact on data collection but only on the playback. You can always request CPU
detail data not be generated on playback but will now also have to request interrupts
not be included as well by collectl -s-jC....
Why can't I see Process I/O statistics?
You need to be running version 2.6.22 of the kernel or greater and it must have
process I/O statistics enabled. The easiest way to check is to see if /proc/self/io
exists. If not, you don't have them enabled and will need to rebuild your kernel and the
instructions for doing so are beyond the scope of this FAQ. If you do rebuild, make sure
you have the following symbols enabled: CONFIG_TASKSTATS, CONFIG_TASK_XACCT and
CONFIG_TASK_IO_ACCOUNTING.
I'm getting an error that formatit.ph can't be found
This component of collectl must be in the same directory as collectl itself. On startup
collectl looks at the command used to start it and from there determines its location by
following as many links as may be associated with that command. It then extracts its
directory name from the last link (if any) in the chain. If one has set up a set of links
such that the last one uses a relative path, when collectl prepends that path to formatit.ph
it's likely not to find it and hence this message. To fix the problem
simply specify the complete path in the final link.
When I use an interval >4 seconds I'm getting non-uniform sample times
Awhile back I found a problem on a SuSE 10 system that was running with a new version
of glibc that changed the granularity of timers from micro-seconds to nanoseconds and
therefore went from 32 to 64 bits. Guess what, 4.3 seconds is > 32 bits! Once I
reported this to the author of HiRes he immediately (within hours) release version
1.91 which addressed the problem. A newer version of HiRes should be the remedy.
I'm getting settimer messages on the console and in dmesg
This problem is actually another form of the previous one and is related to version 2.5 of
glibc. See more details on what this means and how to correct it here.
Why is a set of process data missing during playback?
This actually applies to slab data too. Both types of data contain counters and in order to
display them as rates, collectl needs to read a sample as a baselevel - that's why you see the
waiting for n second sample message when collectl first starts. This applies to
playback too and that's why when you playback data from a file you never see the first
sample. Since process/slab data are usually collected at a different frequency, collectl
has to read more intervals to get to the proper baselevel. As an example, consider data collected
at 00:01AM. When you play it back, collectl is able to include processes/slabs in its baselevel
since this data is typically collected on an intergral minute boundary. It will therefore start playing back non-process
data at 00:01:10 and process data at 00:02:00. In fact, if you play back data from 00:01:10
collectl reads data from the previous interval and will still report process data for 00:02:00.
However if you play back data from 00:01:20 it will set its baselevel to 00:01:10 which does not
contain process data and so will not report process data until 00:03:00.
Therefore, if you're not seeing process data for the time interval you've chosen use an earlier
value for --from.
I played back all summary data with -P but got a cpu detail file. Why?
This will happen when you try to convert interrupt data to plot format file. In fact if you include
-m you will see a message that this is automatically being done for you. The basic problem is that by
definition summary data is written in a fixed format to help ease the automated processing of the tab
file. However interrupt data by definition is variable since there is a counter for each CPU. By
forcing a CPU detail file, there is now a place to put the interrupt data. To get rid of the message
simply include cpu detail when in playback mode.
What does this message mean: -sj or -sJ with -P also requires CPU details so adding -sC
This message is displayed whenever you try to convert interrupt data to a file in plot format and have not included
CPU detail data as explained in the previous section. This message is only displayed when -m is included.
Why are some process usernames being reported as numbers?
Collectl uses the usernames from /etc/passwd to associate with a process's UID. If it can't find that UID in
the file there is no username to associated with it and so collectl
will report the UID instead. This can happen when using NIS in which case username information is stored
remotely OR if playing back a file on a different host than where the raw data was collected. To get around this
problem, copy (or create) the passwd file with the correct relationships to the local system and point
collectl to it with --passwd OR change the Passwd entry in collectl.conf to point to that file if
you get tired of using --passwd.
Why don't I get a prc file with -P and --rawtoo?
The data in the raw file is essentially identical to that in the prc file so the theory is why repeat it
when you can easily create it from the raw file. Furthermore the raw file is already compressed and so
is a lot smaller.
I'm getting an error from RPM: perl >= 0:5.008000 is needed by collectl...
This seems to be a problem with some older versions of RPM. Include --nodeps to
override the dependency check and collectl will install and run just fine.
What does this mean: New network device found: xxx?
When collectl was first developed, the assumption had been that to add new devices, especially a
network device, one would have to power off the machine and reboot, making the possibility of a
new device showing up after boot unlikely. Therefore, when collectl first starts up it identifies
all the networks on the system and if logging to a file writes this information into the header
so that during playback it can all the network names without having to look into the file. If a
new network device does becomes visible after collectl started or during playback, you'll see this message.
However, this message has recently been seen when running with InfiniBand, so a devices obviously
showed up after collectl started. This message also means that the new device has been added to the
end of the list of known networks. If it in fact should have been inserted in the middle of that
list the results are unpredictable and will probably be wrong. Be sure to let me know if this occurs.
Clearly as newer versions of linux/hardware emerge, the types of hardware changes that can occur
without shutting down the systems will no doubt increase and so perhaps more of these types of
situations will occur.
What does this mean: NumNets in header is 'x' but only 'y' listed...?
This message only occurs during playback. Versions of collectl older than 3.5.1 were
updating the internal number of network devices when a new one was encountered in real-time or during
playback, but they were forgetting to update the names used the header as well. This meant when a
new logfile was created after midnight (or whenever it was rotated), the new value for the number of
networks was changed in the header but not their names. If you see this message it's simply pointing
out that fact and you WILL later see New network device found: xxx messages as those new
networks are found in the raw file during playback. This should not occur with raw files created after
V3.5.1.
Why is -sC showing CPUs with no load and no idle?
Somewhere along the kernel started incorrectly updating the CPU usage stats in /proc/stat and if that's
the case with your kernel, you'll see this behavior. There is really nothing collectl can do about it
as it relies on the kernel for accurate stats. Furthermore, before you run out and try a different tool
remember that all tools use /proc/stat to report CPU usage and so that won't help.
When did playing back a file start taking so long?
This problem was first discovered when a user upgraded to RHEL6.2, but it turns out it has nothing to do
with that distro but rather the underlying compression library that collectl relies on, namely Compress::Zlib.
It turns out when that compression library went from 1.42 (last rev of the 1.0 architecure) this broke so
any versions in the 2.0 generation will show this problem. You can see which version of zlib you're using
by running 'collectl -v'. If the playback times are too long for you, just do what I've been doing for
years now and run collectl with -P --rawtoo and you're get both a raw file as well as a plot file with
not a lot of extra overhead.
What does the message -D requires -f OR -A server mean?
First of all, this should only be seen by someone who has modified /etc/collectl.conf because it already
specifies -f. The point of the message is that when you run collectl as a deamon, it needs to know what
to do with its output since there is no terminal available. That destination must be a file OR collectl
must be run as a server, sending its output out over a socket.
This module reports on several variables that are not collectled as part of collectl's core
metrics, partly because some of them don't exactly fit into collectl's main stats, but some users
still find useful. There are also a few instructive techniques used in this simple module that
are worth calling out:
use of its own routine grepProc() to find a specific entry in /proc and return either the first
instance or the count of instances, noting that using the real grep has far too much overhead
due to the grep process start-up and tear-down
support for a different monitoring interval yet still displays, exports and logs the data every cycle
ability to pass additional parameters with values, in this case the interval
use of getExec() for running another process to get the number of logged in users, noting this is
slower than getProc() and the command being used is slow enough that a different monitoring interval
is needed to avoid excess overhead.
The following example shows one importing both hello.ph and misc.ph while displaying cpu data and
running all at the same interval:
In this example we're just doing the 2 imports, setting the misc monitoring interval to 2
and exporting the data with lexpr. As you can see, the hello data is reported every interval
but the misc data only every other one:
collectl-3.6.9/docs/slub.jpg 0000644 0001750 0001750 00000033576 12230241726 014112 0 ustar mjs mjs JFIF x x C
$.' ",# (7),01444 '9=82<.342 C
2! !22222222222222222222222222222222222222222222222222 J"
} !1AQa"q2#BR$3br
%&'()*456789:CDEFGHIJSTUVWXYZcdefghijstuvwxyz
w !1AQaq"2B #3Rbr
$4%&'()*56789:CDEFGHIJSTUVWXYZcdefghijstuvwxyz ? (
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
Mr]u)@Ӭ zd "?L+]EhWZ[t&J7}wOY,QG2IG t ?Ze>t%%kf\DS w]n|Wh3Z2Y
2y 2߷<⼪
ZA{SkS=B-;;1~xs e|%31ﰿع" pU"'h@Oqpo3HK(\ 9>KZe>t%%kf\DS wU4 2 ~Q_jDM+JL
ޠH/|
mjgEtӇp5 {c }J~F 1R{+6 XF?ޢ:密|-qEarmo.(9 w.FG=nu=z|$6Cˏ}P &bVhYx?X|ΏxvJ.&*Gr3@-|Ix*9ad}0 }+|_ygͤhI4%v®H] wzƫxgj-C{{{{@.`,_< BT Xqlyr {P> NLt6xl\Ze!R3px xT6-;"KgHՋ$j˅g 3qG5h [^x5F t+˅4p;@+ `1]<{/ᾏ i
ͮId;0pypF; t>%<6/|AJ,ٰy" / k"
Utun 5e6ʐ iAr}OZt]"ß |e3[RFQ*g%=n@ֵǭj:xI]f["˺#!8G@pA +W_tOIGmIiRPѴN ;}*8>l羿* V< $\Ƣ_XD&a6 |Ò{V#|QHS)kVKP 9Dq } ĞBI,9 X\a9:x3Zw+-×Z$ʩ2A Qc'qKi7?k*Ė7
y$dZOh4յ]=-쑽ԫ2f$qAsc\+&)ΟrBHD2 #0 jŗ^Zn.0`!Gq -⧉mdv;7̕!dڲ<8oz ր7Zi~=Ԯ BI5xPO
M_G$`q ,.0 Z% ?6DO\"Z ǀN@d--|Do[h&0 *4jq䑒p v^>?'?xbSS˺HQ${fpcՉ@6 ]4͡JWPk#a ;q9* FM_r7t`h[-o +x1B lm~ZKXmN' I,I0`el~^jy};#0$6ǕnbGo kBA@w6[Gm~m!n~>h{md Grۈ9}fܿӲ1ETTmZa \ Cۍw6?-h3[n
,J!QT ``;q E >S_iQqSIxGFc`(6uq ~!S_i6>γFQ!O/$ $u;a'89IvwΡIwR3F?=?RG?v3ϧŇIť_ʹ>Ig;#=3ާo>mSY}?N|Qn]~mJ Þp.>S+L_W }gPL|:Wɤ4; Yv:}\WI2E)X+ #֮xz_|ioI;X(mY q9Ksk4 SL}(lyj{ î|{3 a'89RϧŇIť_ʹ>Ig;#=3ޕwZ6}9۶oPq!ͪk/ϊ-K܍CsqX NCŞcߥEt$R2 ,
B=hr^xG_/+ 3[>\c N}bYR^iX$qfc5 [N
,hhf07qsp+g@9\c&O+I|6@s{И8<7"~o w9ݷ ;Wg A иxPã2
s1=2k=1~4- mE<2ܪr^kC7~:KM/1f0.ӎ(BvѭES$( n?[WםxD} -k9T6aETQ@Q@Q@Q@Q@Q@Q@qWP/Y۪kT𦓫 oKŹ1,%܊X"7Pp]iٓ8F}/ z ? " C ;G z 9>U:wcԤ<=n5I,!#B
&Ӕ$ p9=? ?wy hCTq1xB&{iuzݢHѣƀ OSeׄE|0vP* 3| qc#Ӛ c wv2 s_ z6RÞn4\%WBɜo\ˆ_qC t r/ z ? /`QR X =? stǩIx{Q,jYBF'M(H#o5"OG. ${[~9r9M#× .}yj,PQhN>bI=s]G z ? ̅-t?vh~eC
5j8QxWMhiY "X$kW=? ?wy h@" C ;G z |܊=? ?wy h` G q ڽy x~!uIaJ^ĕr<֝mAIt~Rw _GnؽX;3@8Z++
lr@N
zne[{H]b# ;K7IQE QE QE QE QE QE QE QE QE QE ɢY&.TrA(u$ XD5i,u c#qBzC*OsFII28##ȧ}Α
Ʃo0\ YS$u9V q
+:Qk۫{=R w"@ x;*ުG ֍ QE QE QE QE QE QE QE QE QE QE QE QE QE QE QE QE QE QE QE QE Uխ t@ZA*y$'Uou)"]9Mtrun9 u w[7%b&rol *Sl)"T}VYg]<)X㈗n3?_ Iop1 jA$ɧ=7I{o֑[DN 7rǫ s\ ( ( ( ( ( ( ( ( ( ( ( ( ( dq 4i$R)WGPUQY-cqi+h,rn, >Cccہ٢+{c"%ߔ5:I$x#=25bW67=oR>h| =zmj|JE<*E^IFNs@R;IQ%+PEM@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@\(6wE@@:\%?j}][$&G7yvk[ǟ,' ,ncVЂu{x)Q85?&w?)k ("2LR? $Q
Z w?_}g|͛:\ }[tr!iZ
Z ?3 K\ Tno,`N&>u* /N[~"Ƒ.&A3GsIQȃO
Z ?3 K\ TmC[C/+]s{lޡ =qWhA@?g|+I]7ZpX)ϟJ7 3 K\ X<:|OP UC՛X]oWv_G,U,Ao nexv\A'*V 'Px5ow9P][l*X ũɢxi_4(7dsۨ7HѾ!gOeI Y;+H k(mX
Wm_?ϗ 69yOTdbx;D} -kn< "> `B
"Ǻƽul5 q@e2Ř{UԶ
q5zU%7-8RNۼ|CkӖ;Y.-us :/ G}~k}b)Coh dQ] " U輋F1QT '&_krICeդF?R۲4k-X-R'h g!Xg8$8547]]D^UCMq|/
[~ @} ι? >c*At4l sB?vqp[\=f
f}kW61\N 1 JDo-ksYgm) 1[b9bI$& !
I \ $ IqMg!ajڦ~&Qn]|ljP WK\4NLx-2I$n7)< q494 &a=NOK&H+ d X>zKH-yGY"%d;`dM'FbSnI)JYpO AtOIR$HVYzm!c jgW#еWQknrFUf:`vX - ?h GIZ7_!!'Ovd1 j<
xg=qoohI(wH0HmrCG? _# *GmS= UQYaEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPvůiOa4۫I,w"Ȥn >7!?6 몢 \ vG!?6 몢rG s Cn ~ C.m?ݟ #UEa ǟ|&bV.3\fl`8 ks Cn ~ (X+ e
z? \ v]U]${ r[V.%"wiʡb F8.ߟҧ 2s =uTQv r\ ۮ?G˟ug QErG s Cn ~ jXӧQk: ]9V kc~ ţGcY;c+0y?֭ s Cn ~ (H9_C.m?ݟ # e
zꨢ9# }XMj0d; \ ۮ?G.Ò=W˟ug s Cn ~ (H9_C.m?ݟ #IM?__S\:ڱPWQErDZ s Cn ~ C.m?ݟ #UEa + e
z 5qC,'m3 ßZh9# \ vG!?6 몢c 2s = \ ۮ?G.Ò=W˟ug s Cn ~ (H9_C.m?ݟ # e
zꨢ9# \ vZ
ڭw hk/qmT
j֢
(((((((((((((((((((((((((((((((((((((((((((( collectl-3.6.9/docs/Graphite.html 0000664 0001750 0001750 00000013630 12230241727 015064 0 ustar mjs mjs
Exporting Data to Graphite
Exporting Data to Graphite
Introduction
With the release of Collectl Version 3.6.1, you can now send collectl data directly to
graphite . For existing collectl users this now provides you with yet another way to store/plot collectl data, whether on
a single system or hundreds. For graphite users who are not yet collectl users, you now have access to literally hundreds
of performance metrics:
Since all collectl instances collect this data at the same time, system noise on
clusters running fine-grained parallel jobs is reduced, though for larger clusters.
You can still log all the data collectl collects locally and only send a subset
to graphite, reducing the load on both graphite and your network.
You can monitor as infrequently as you like and send data to graphite at a
coarser frequency of either of the average, minimum or maximum values over that interval.
The r= switch, something unique to the graphite plugin, can help reduce the instantaneous
load on the graphite server itself.
All this at collectl's low monitoring overhead
Usage
You use this export like any other, the only required option being the address to send the data to as in
the following example:
collectl --export graphite,192.168.1.113
However you should also note that since by design this export does not provide any terminal output, there
are only 2 real ways to make sure it is doing what you expect, the first being to inspect graphite's
whisper storage area for your particular host name and make sure the data you're collecting is in fact
showing up there:
ls /opt/graphite/storage/whisper/poker
cpuload cputotals ctxint disktotals nettotals
or to simply run with the debug mask set to 1, which tells the graphite module to echo all the data it is
sending to graphite, noting in this case even though collectl is collecting cpu, disk and network data we're
not sending cpu data to graphite. This is something you might do if logging more data to disk than you are
sending to graphite, which in this case we are:
tip - if you add 8 to the debug flag, eg d=9, this tells the graphite module not to
actually establish the connection with graphite's carbon listener but to only echo the data that would
have been sent.
Once you're happy with the switch settings, be sure to update the DaemonCommands in /etc/collectl.conf
and restart the collectl daemon to make them take effect.
Switches unique to graphite
e=escape When sending data to graphite, collectl prefaces each line item with the hostname. If that name includes
a domain name, extra dots add additional levels the the variable names which may not be desireable. By including
an escape character, those dots will be replaced by that character.
r=seconds
By design, collectl calls the export module as soon as the required data has been collected and collection
is synchronized to the nearest milli-seconds across a cluster, this means all instances of collectl will send
their data to graphite at almost exactly the same time. This high burst of data can overwhelm graphite and
so to reduce the load when that is found to be a problem, OR if you just want to smooth out the load you can
use r=seconds which literally means delay sending your data to ganglia by a random number of micro-seconds
<= seconds.
There is an additional caveat and that is that this stall must have completed by the end of the
current data collection periods and so you're restricted to a maximum delay of the interval less 1 second.
This means if you run collectl with -i1, you can't use -r. However, since most users run collectl with
intervals of 5 or 10 seconds, values of 4 or 9 should be more than sufficient. And if you choose a collection
interval of 30 seconds you may still want to use a value of r closer to 5 or 10 seconds so that the data will
arrive at graphite reasonablly close together.
For help with what other valid switches are, you can actually get the graphite module itself to tell
you like this:
collectl --export graphite,h
Communications
Collectl will attempt to establish a TCP connection to the specified address/port, noting the default port is 2003.
If that connection cannot be established, collectl will report an error but not exit! This is because
graphite itself may be down and need to be restarted.
collectl --export graphite,192.168.1.113,d=1,s=dn
Could not create socket to 192.168.1.113:2003. Reason: Connection refused
By design when collectl assumes the graphite address is correct and will try to reconnect every monitoring
interval. Further, to avoid generating too many errors, it will silently continue to retry and only report
the connection failure every 100 times, a constant you can modify in the graphite.ph header if you really
care. Once graphite comes back online collectl will again start sending data to it.
By design, collectl gathers more data than is possible to display in an efficient, easy to read,
compact form. However, most user want their data displayed in such a form for easy
interpretation. Therefore, collectl will attempt to display all data in a single line, often
choosing a subset of the complete data for each subsystem. If the user has selected too
many systems, each line may exceed the display width and wrap. When this happens either make the
terminal window wider (maybe even using a smaller font) or choose less subsystems.
This is referred to as brief format and is collectl's display format of choice and
therefore the default. Verbose mode displays more information and results in multiple
lines of output.
Collectl will try its best to select a format consistent with the user's selection
criteria, using brief mode whenever possible unless explicitly told no to do so.
However there are several instances
when this mode doesn't make sense. For example, detail data will always be displayed in
verbose mode since it takes multiple lines for each sample. When this occurs, collectl
will automatically use verbose which can also be manually forced for non-detail
data using --verbose.
One should note that these formats are not just for interactive use and can also applied
to playback mode as well.
An additional feature of brief output is subtotal mode. If one hits the
enter key at
any time, the next line of output will be the subtotals (or averages on non-counters)
of all columns since
the start of collectl OR the last time the counters were zeroed. To zero the
counters enter Z followed by a carriage return.
Furthermore, if you type A followed by the enter key, the averages
will be reported. The averages/totals can also be displayed during playback
in brief mode by specifying -oA.
To get a better idea of what the output actually looks like, see the examples.
There are several switches that provide even more control over the look of the output in addition
to --verbose as described above. They are:
--home moves the cursor to the home position before displaying verbose output at the start
of each interval. Only available in interactive mode, this results in a look-and-feel similar
to the top command.
--procfilt and --slabfilt effect the output format for those respective output formats
in that these typically cause a much smaller number of processes or slabs (if used in
conjunction with --slabopts S) to be displayed, sometimes as little as a line
or two and it was felt repeating the interval header when only processes or slabs are the only data
being reporting, was too distracting and so it left off. Be sure to try it with -oT for better
clarity.
--top is very similar to the linux top command in that it shows a small subset of processes
sorted by the top consumers of the cpu, I/O or even page faults. You can even use -s to add
subsystems to the display in brief or verbose mode as well. By default this format sorts by
the top CPU users but you can choose virtually field. If you choose one of the slab field
names it will show the top slabs sorted by that field name.
The best way to really understand how these work in conjunction with each other is to try them
out. And don't forget you can use --top with playback too!
This table shows the number of memory fragments by pagesize in increasing powers of 2 for various types of
memory defined by the combination of Node and Zone.
CPU Data, collectl -sC
# SINGLE CPU STATISTICS
# CPU USER NICE SYS WAIT IRQ SOFT STEAL IDLE INTRPT
These are the same fields as reported for verbose CPU Summary, each preceeded by the
CPU number. If running collectl V2.5.0 or greater AND you request interrupt summary
data, the INTRPT field will also be included.
CPU
The CPU number which the stats are associated with
User
Time spent in User mode, not including time spend in "nice" mode.
Nice
Time spent in Nice mode, that is lower priority as adjusted by
the nice command and have the "N" status flag set when examined with "ps".
Sys
This is time spent in "pure" system time.
Wait
Also known as "iowait", this is the time the CPU was idle during an
outstanding disk I/O request. This is not considered to be part of the total or
system times reported in brief mode.
Irq
Time spent processing interrupts and also considered to be part of
the summary system time reported in "brief" mode.
Soft
Time spent processing soft interrupts and also considered to be part
of the summary system time reported in "brief" mode.
Steal
Time spend in involuntary wait state while the hypervisor was servicing
another virtual processor.
Intrpt
If the interrupt summary stats were requested at the same time, this
will be included which is the aggregate number of interrupts for each CPU.
Disk Data, collectl -sD
If you specify filtering with --dskfilt, the disk names that match the pattern(s)
will either be included or excluded from the the summary data. However, the data will
still be collected so if recorded to a file can later be viewed. Note: if you
specify --dskopts f, fractional values will be reported for some of the fields
for more precision.
Name of the disk the statistics are being reported for.
KBytes
KB read/sec
Merged
Read requests merged per second when being dequeued.
IOs
Number of reads/sec
Size
Average read I/O size in KBytes
KBytes
KB written/sec
Merged
Write requests merged per second when being dequeued.
IOs
Number of writes/sec
Size
Average write I/O size in KBytes
RWSize
Average combined read and write I/O size in KBytes. This is not the average
of the read and write sizes but rather the sum of the reads/write divided by the number
of I/Os
QLen
Average number of requests queued
Wait
Average time in msec for a request has been waiting in the queue
SvcTim
Average time in msec for a request to be serviced by the device
Util
Percentage of CPU time during which I/O requests were issued
Count of current errors. Since these are typically infrequent, it
is felt that reporting them as a rate would result in either not seeing them OR
round-off hiding their values.
Interrupts, collectl -sJ
# INTERRUPT DETAILS
# Int Cpu0 [Cpu...] Type Device(s)
Int
Interrupt number within the range 0-255. Note that
only those interrupts that have had any activity since the last monitoring
interval will be reported
CPUn...
The CPU for which the interrupt count is being
reported. There will be one column/CPU
Type
Interrupt type
Device
The names of the devices which are
generating this interrut
Lustre Data, collectl -sL
There are several formats the lustre detail data can take based on whether you're
looking at a client or an OSS (there is not any MDS specific detail data, though it does
share the same disk-level buffer size data as the OSS). Furthermore, if one specifies
the -sLL form of the detail switch OST level details will be reported where
appropriate.
This form also includes the reads/writs within the filesystem, but also add the sizes of
the RPM buffers. Since these numbers always apply to OSTs you need to use the -sLL
form of the subsystem switch.
Filsys
Name of the filesystem these stats apply to
Ost
OST name within the filesystem
RdK
KBs read/sec
Rds
Reads/sec
nK
Number of pages of of this size read
WrtK
KBs written/sec
Wrts
Writes/sec
nK
Number of pages of of this size written
Lustre Client Metadata, collectl -sL --lustopts M
# LUSTRE CLIENT DETAIL: METADATA
#Filsys KBRead Reads KBWrite Writes Open Close GAttr SAttr Seek Fsync DrtHit DrtMis
Lustre OSS and MDS Disk Buffers, collectl -sL --lustopts D
This display is very similar the the RPC buffers in that the sizes of different size
I/O requests are reported. In this case there are requests send to the disk driver.
Note that this report is only available for HP's SFS.
This is also known as numa
data and provides detail information about memory utilization in each numa node.
# MEMORY STATISTICS
# Node Total Used Free Slab Mapped Anon Locked Inact Hit%
Node
Numa node number, which is usually the same as a physical socket
Total
Total physical memory
Used
Used physical memory. This does not include memory used by the kernel itself.
Free
Unallocated memory
Slab
Memory used for slabs, see collectl -sY
Mapped
Memory mapped by processes
Anon
Anonymous memory
Locked
Locked memory
Inactive
Inactive pages, which is the sum of Inactive(anon) and Inactive(file).
Note that Inactive(anon) is not considered nor included in the previous
anonynous memory field.
Hit%
It is currenlty not entirely clear how useful this number actually is as it refers to the hit percentages
for both local and foreign memory as a single number. Most importat, it does not refer to memory access
but rather memory allocation. In other words, it does not differentiate between one failing to allocation
memory and only referencing it a small number of time vs a very large number of times. Clearly the latter would
be more interesting from a performance perspective.
Network Data, collectl -sN
If you specify filtering with --netfilt, the names that match the pattern(s)
will either be included or excluded from the the summary data. However, the data will
still be collected so if recorded to a file can later be viewed.
# NETWORK STATISTICS (/sec)
#Num Name KBIn PktIn SizeIn MultI CmpI ErrsI KBOut PktOut SizeO CmpO ErrsO
Num
Each network interface is numbered, starting with 0
Name
Name of the interface
KBIn
Incoming KB/sec
PktIn
Incoming packets/sec
SizeI
Average incoming packet size in bytes
MultI
Incoming multicast packets/sec
CmpI
Incoming compressed packets/sec
ErrsI
Total incoming errors/sec. This is an aggregration of incoming errors. To see explicit
error counters use --netopts e
KBOut
Outgoing KB/sec
PktOut
Outgoing packets/sec
SizeO
Average outgoing packet size in bytes
CmpO
Outgoing compressed packets/sec
ErrsO
Total outgoing errors/sec. This is an aggregation of outgoing errors. To see explicit
error counters use --netopts e
Each network interface is numbered, starting with 0
Name
Name of the interface
ErrIn
Receive errors/sec detected by the device driver
DropIn
Receive packets dropped/sec
FifoIn
Receive packet FIFO buffer errors/sec
FrameIn
Receive packet framing errors/sec
ErrOut
Transmit errors/sec detected by the device driver
DropOut
Transmit packets dropped/sec
FifoOut
Transmit packet FIFO buffer errors/sec
CollOut
Transmit collisions/sec detected on the interface
CarrOut
Transmit packet carrier loss errors detected/sec
NFS Data, collectl -sF
By default, collectl will report all 6 types of nfs data (clients and servers for
all 3 versions of nfs) unless one has limited the reporting with --nfsfilt.
For versions prior to 3.2.1 which only collected a single type of
data, collectl will only report that single type.
Also note that nfs V2 records 2 fields that V3 doesn't record and V4 records many
more. At this time and detail data is standardized on the V3 format and fields
not collected will be left blank. These fields map onto those reported by
nfsstat as indicated and are reported as rates.
There are actually multiple formats process data can be displayed in, the default being the one
shown immediately below. By using --procopts as shown in later examples, you can change
what is displayed, noting that when playing back the data from a raw file, all information
has been recorded and so you can actually play it back multiple times and see different views.
These switched can also be use in conjunction with --top.
# PROCESS SUMMARY (faults are /sec)
# PID User PR PPID S VSZ RSS CP SysT UsrT Pct AccuTime RKB WKB MajF MinF Command
PID
Pid of the process
User
Name of user which this process is running under. In playback mode on a
different machine, use -oP to direct collectl to use the password file named in
collectl.conf (default is /etc/passwd) to lookup the corresponding username.
Otherwise the UID will be reported instead.
PR
Process priority
PPID
PID of this process's parent
S
Process State: S - Sleeping, D - Uninterruptable Sleep, R - Running, Z - Zombie or T - Stopped/Traced
VSZ
This is the amount of VS memory used by this process
RSS
This is the amount of RSS memory used by this process
CP
CPU number this process is currently running on
SysT
The amount of System Time this process used during this interval
UsrT
The amount of User Time this process used during this interval
Pct
Percentage of the current interval taken up by this task (the User and System
time are used for this calculation)
AccuTime
Total accumulated System and User time since the process began execution
RKB
This is the number of kilobytes of data written by each process. Both
this and the WKB field are only present if the kernel had proces I/O
monitoring enabled which is not the default as of 2.6.23.
WKB
This is the number of kilobytes of data read by each process
MajF
Major Page Faults per second
MinF
Minor Page Faults per second
Command
Command that is running. Path and command line options are NOT included
unless --procopts w
Process Data, collectl -sZ --procopts x
This format is essentially idential to the last except that it adds extended information
to the display, specifically VCtx and NCtx.
# PROCESS SUMMARY (counters are /sec)
# PID User PR PPID THRD S VSZ RSS CP SysT UsrT Pct AccuTime RKB WKB VCtx NCtx MajF MinF Command
VCtx
Voluntary context switches
NCtx
Non-voluntary context switches
Process I/O Data, collectl --procopts i
# PID User PPID S SysT UsrT Pct AccuTime RKB WKB RKBC WKBC RSys WSys Cncl Command
PID
Pid of the process
User
Name of user which this process is running under. In playback mode on a
different machine, use -oP to direct collectl to use the password file named in
collectl.conf (default is /etc/passwd) to lookup the corresponding username.
Otherwise the UID will be reported instead.
PPID
PID of this process's parent
S
Process State: S - Sleeping, D - Uninterruptable Sleep, R - Running, Z - Zombie or T - Stopped/Traced
SysT
The amount of System Time this process used during this interval
UsrT
The amount of User Time this process used during this interval
Pct
Percentage of the current interval taken up by this task (the User and System
time are used for this calculation)
AccuTime
Total accumulated System and User time since the process began execution
RKB
Attempt to count the number of bytes which this process really did cause to
be fetched from the storage layer by doing calls to read_bytes.
This is done at the submit_bio() level, so it is accurate for block-backed
filesystems.
WKB
Attempt to count the number of bytes which this process caused to be sent to
the storage layer by doing calls to write_bytes.
This is done at page-dirtying time.
RKBC
Number of bytes which were read via read, readv, pread and sendfile. Since
these requests are satisfied from kernel pagecache they won't be accounted for
by RKB, because they didn't require any I/O.
WKBC
Number of bytes which were written via write, writev, pwrite and sendfile.
Like RKBC, since the I/O uses the pagecache these values won't be accounted
for by WKB.
RSys
Number of read syscalls, specifically: read, pread, readv and sendfile
WSys
Number of write syscalls, specifically: write, pwrite, writev and sendfile
Cncl
Number of cancelled write bytes.
Process Memory Data, collectl --procmem
# PID User S VmSize VmLck VmRSS VmData VmStk VmExe VmLib VmSwp MajF MinF Command
PID
Pid of the process
User
Name of user which this process is running under. In playback mode on a
different machine, use -oP to direct collectl to use the password file named in
collectl.conf (default is /etc/passwd) to lookup the corresponding username.
Otherwise the UID will be reported instead.
S
Process State: S - Sleeping, D - Uninterruptable Sleep, R - Running, Z - Zombie or T - Stopped/Traced
VmSize
Size of Virtual memory used by the entire process
VmLck
Size of Locked Virtual Memory
VmRSS
Size of Resident Virtual Memory
VmData
Size of Virtual Memory used for heap
VmStk
Size of Virtual Memory used for stack
VmExe
Size of Virtual Memory used for exe and statically linked libraries
VmLib
Size of Virtual Memory used for dynamically linked libraries
VmSwp
Size of Virtual Memory used for swapping. This does not necessarily mean a process
is actively swapping but only that the memory has been mapped.
MajF
Major Page Faults per second
MinF
Minor Page Faults per second
Command
Command that is running. Path and command line options are NOT included unless --procopt w
Slab Data, collectl -sY
There are actualy 3 different formats for slab data. The first applies to all kernels prior to 2.6.22
and contains the same fields as the Summary report for each named slab and loses the caches
fields.
Total number of objects that are currently in use.
Bytes
Total size of all the objects in use.
Alloc
Total number of objects that have been allocated but not necessarily in use.
Bytes
Total size of all the allocated objects whether in use or not.
Slab Allocation
InUse
Number of slabs that have at least one active object in them.
Bytes
Total size of all the slabs.
Total
Total number of slabs that have been allocated whether in use or not.
Bytes
Total size of all the slabs that have been allocted whether in use or not.
Diff
Change in size of this slab since last sample in bytes
Pct
Percentage change in size of this slab
This second format applies to the new SLUB allocator starting with the 2.6.22 kernel. As with the old
format slab detail report, the same fields as are found in the Slab Summary Report are shown for each named slab.
# SLAB DETAIL
# <----------- objects --------><--- slabs ---><---------allocated memory-------->
#Slab Name Size /slab In Use Avail SizeK Number UsedK TotalK Change Pct
Objects
Size
Size of a single slab object
/Slab
The number of objecs in a single slab
InUse
The total number of objects that have been allocated to processes.
Avail
The total number of objects that are available in the currently allocated slabs.
This includes those that have already been allocated toprocesses.
Slabs
SizeK
The size of one slab, which typically contains multiple objects
Number
This is the number of individual slabs that have been allocated and
taking physical memory.
Memory
UsedK
Memory used by those objects that have been allocated to processes.
TotalK
Total physical memory allocated to processes. When there is no filtering
in effect, this number will be equal to the Slabs field reported by -sm.
Change
Change in size of this slab since last sample in bytes
Pct
Percentage change in size of this slab
The third format uses a common format based on the slabtop utility for displaying
top slab data sorted by any of the listed column headers. All on need to do is use one of
these names as the argument to the --top switch in lower case (use --showtopopts
for full list of options to the --top switch). All the same rules apply for
controlling the number of lines in the data section, mixing in subsystem data and even using
with playback mode.
# TOP SLABS 15:38:08
#NumObj ActObj ObjSize NumSlab Obj/Slab TotSize TotChg TotPct Name
NumObj
Total number of objects that are available, which includes those in use
ActObj
Number of objects that are in use
ObjSize
Size of an individual slab object
NumSlab
Total number of slabs that have been allocated
Obj/Slab
This is the constant number of objects that fit into one slab
TotSize
Amount of memory consumed by this slab even if not all objects are actualy allocated
TotChg
Change in size of this slab since last sample in bytes
Collectl actually provides 2 different mechanisms for socket communications and depending
on which style you intend to use, you should pick the appropriate one. Regardless of
which style you choose, collectl will send all data that might normally go to a terminal
to that interface, prepended with the hostname for easy recognition at the other end.
One should also note that you can simultaneously record the data locally in a raw
file, a file in plot format or even both depending on your needs.
Short-lived communications
The model for this type of communications is one in which some other tool such as
colmux starts
collectl on one or more remote systems, instructing it to send its output back
to the initiator until the initiator terminates the connection.
At this time collectl will
permanently exist. Using this style the initiator of the communications opens
the socket and then typically starts collectl using ssh and includes its
address/port in the command string using collectl's -A switch. The key things
to note when using this style are:
This mechanism will never survive a reboot without extra work on the initiator's
part which must include restarting collectl
As soon as it starts, collectl will open a socket to the specified address and immediately start sending
its output back to it
If the other end of the socket goes away, collectl will exit
If collectl is unexpectedly terminated (such as sigint or reboot), it will
close the socket without warning the other end and therefore the program at the
other end of the communication path must be prepared to deal with it
Long-lived communications
This mechanism has collectl create the initial socket and listen for a remote connection.
In this mode, collectl continues to collect data and at the beginning of each monitoring
interval looks for a new socket connection request.
If it receives a request, collectl then creates a
connection back to the requestor and immediately starts sending data to it.
If that connection terminates
collectl closes down its side and then goes back to listening for a new connection.
The key points about this style are:
This style of communications will survive a reboot if configured in collectl.conf
The switch to put collectl into this mode is also -A but rather than supplying an
address as the argument use the text server
The other end of the connection needs to periodically try to connect and if it fails
must try again. If that connection exits, it must clean up its end and go back to
trying to reconnect again.
If you want collectl to terminate you much manually do so, noting after a reboot it
will restart.
To verify collectl is functioning as expected, run the utility
/usr/share/collectl/util/client.pl passing it the address of the system collectl is
running on. It will display anything collectl sends to it over the socket.
Collectl has the ability to monitor processes in pretty much the same way as
ps or top do as can be see here:
# collectl -sZ
# PROCESS SUMMARY (faults are /sec)
# PID User PR PPID S VSZ RSS SysT UsrT Pct AccuTime MajF MinF Command
21502 root 15 1749 S 6M 2M 0.00 0.00 0 0:06.40 0 0 /usr/sbin/sshd
21504 root 15 21502 S 4M 1M 0.00 0.00 0 0:00.79 0 0 -bash
22984 root 15 1 S 7M 1M 0.00 0.00 0 0:00.78 0 0 cupsd
23073 apache 15 1914 S 18M 8M 0.00 0.00 0 0:00.01 0 0 /usr/sbin/httpd
You can select processes to monitor by pid, parent, owner or command name (see
section on Filters below).
When using names, you can use partial or full names or even use strings that
were part of the command invocation string such as parameters.
The main benefit of monitoring processes with
collectl is that you can coordinate the sample times of process data with
any of the other subsystems collectl can monitor.
The way you tell collectl to monitor processes is to specify the Z subsystem
and any optional parameters with --procopts. Since
monitoring processes is a heavier-weight function, it is recommended to use a
different interval, which can be specified after the main monitoring interval
separated by a colon. The default is 60 seconds. Therefore, to monitor all
the processes once every 20 seconds and the rest of the parameters every 5
simply say:
collectl -sZ -i5:20
The biggest mistake people make when running this command interactively is to
leave off the interval or specificy something like -i1 and not see any process
data. That is because the default interval is 60 seconds and they just
haven't waited long enough for the output! This should obvious since collectl
will announce it is waiting for a 60 second sample.
There are also a few restrictions to the way these intervals are specified. The
process interval must be a multiple of the main interval AND cannot be less
than it. If you specify a process interval without a main interval, the main
interval defaults to the process interval.
Finally, as with other data collected by collectl, you can play back process
data by specifying -p. While not exactly plottable data, you can specify
-P and the output will be written to a separate file as time stamped space
delimited data, one process per line.
Options
As stated earlier, you can specify options specific to process monitoring. These
apply to all forms of process output unless otherwise noted, including --top
and --procanalyze
c: include cpu times of children who have exited with their parent
f: use cumulative totals for maj/min page faults
i: Show alternate format which includes all io counters
m: Show alternate format which includes all memory sizes
p: Never look for new pids or threads after startup (improves performance)
r: Only show root name of command, leaving off path
t: Look for thread for all processes
w: Include command arguments, making a wider display. Can be combined with r.
z: Only show processes with non-zero sort field. This only applies to --top.
Filters
You can tell collectl to monitor a subset of processes by using the
--procfilt switch followed by one or more
process selectors, separated by commas (see collectl -x to see a
detailed list). For the most part, the use of filters is pretty
straighforward in that if you want to see all processes whose parent is 1234
as well as those that contain http in their command name, you would
specify a filter of --procfilt P1234,chttp. However there is one
important distinction to keep in mind. The c prefix says to select on the
command name only whereas the f prefix says to look at the entire command
string including arguments. In other words, if you're editing the file
abc and try to select it via --procfilt cabc you'll never see it.
This is particularly annoying when monitoring perl scripts since the name of
the command is perl and the name of the script shows up as an argument.
If a plus sign immediately follows a
process selector any processes selected by it will have their threads
monitored as well. See collectl -x or man collectl for more details.
Dynamic Process/Thread Monitoring
A unique feature of process monitoring is that processes specified with a
selection list via --procfilt do
not have to exist at the time collectl is run. In other words, collectl will
continue to look for new processes that match this selection list during every
collection cycle! While this is indeed a good thing if that is
what you want to do, it does
come with a price in overhead: not a lot, but overhead never-the-less.
If you do not want this effect and only want to look at those processes that match
the selection list at the time collectl is started, specify --procopts p
to suppress dynamic process discovery.
This holds for process threads as well, suppressing looking for new ones.
Perhaps the best way to see this in effect is to run collectl with the
following command:
collectl -i:.1 -sZ --procfilt fabc
The .1 for an interval is not a mistake. It is there to show that you
can indeed use collectl to spot the appearance of short
lived processes - just don't do it unless you really need to. The --procfilt
switch is saying to look for any processes invoked with a command that contains the
string 'abc' in it. When this command is invoked there shouldn't be any output unless someone IS
running a command with 'abc' in it. Now go to a
different window or terminal and edit the file abc with your favorite editor.
You will immediately see
collectl display process information for your editor and when you exit the editor the output will stop.
The Time Fields
The SysT and UsrT represent the system and user time the line item spent during
the current interval. One might think this means that in a 60 second interval the
most time a process could spend is 60 seconds. Not quite! If this is a
multi-processor/multi-core system the process could actually spend up to 60 seconds
on each core, so just be careful how the times are interpretted. The Pct field is
the percentage of the current interval the process had consumed in system and user
time, which can also exceed 100% in multi-processor situations. Finally, since the
AccuTime field accumulates these times it can exceed the actual wall clock time.
When run in non-threaded mode, the times reported include all time consumed by
all threads. When run in threaded mode, times are reported for indivual
threads as well as the main process. In other words, if a
process's only job is to start threads, it will typically show times of 0. If
you rerun collectl in non-threaded mode you will see it report aggregated
times.
Process Memory Utilization
The types of memory utilization displayed as part of the process monitoring output
are the Virtual and Resident sizes. However there are additional
type of memory that collectl tracks and to see them as well as page faults
you can select alternate process display format as follows:
As of collectl Version 2.4.0, if process I/O stats have
been built into the kernel collectl will add 2 additional columns to the
process display named RKB and WKB,
noting in the following example I've set the display interval to
1 second and removed the initialization message from the output. As with all
fields reported as rates/sec these will show consistent values independent of
the interval and if you want the unnormalized value be sure to include
that option with the -o switch as -on.
A particularly useful feature I've found is monitoring one or more processes by name (you
can also monitor by pid, ppid and uid) to see
what they're doing. In this case I'm using the dt program to write a large file and telling
collectl to display any process whose command string matches dt as well as to include
time stamps.
Finally, note that there is more process I/O data available but I chose to leave it off
the default display and instead have the following alternate format. This is
the same methodology used for reporting process memory utilitation, namely you only
see VSZ and RSS in the default display but much more with --procopts m.
Also note in this caseI chose 1/2 second monitoring as well as showing time in msec resolution:
A feature that has been in collectl for awhile has been the --top switch which generates
data in a format similar to the linux top command, though it was limited to process data
only. However, since the inclusion of process I/O statistics as well as the inclusion of a few
additional handy switches, this command has become much more useful as explained below.
In its simplest form, this switch tells collectl to simply display the top consumers of cpu.
However, as of collectl V2.6.4 you can now now tell it to optionally display the list sorted by
I/O or page faults.
Here I'm simply looking for the top processes sorted by page faults with the
command collectl --top flt and the display fills my window, which in this case is only
10 lines high. To look at the top consumers of I/O, simply use --top io instead:
As discussed earlier, threads can be considered for displaying
with --procopts t
which requests all selected processes be examined for threads. You can also
specify a subset of processes be examined by specifying a + with
--procfilt but that's getting into more advanced concepts.
Fnally, in the spirit of saving screen real estate collectl doesn't include command arguments in
the output but including --procopts w will request a wider display that does include
them. In fact you can get an even narrow display by including --procopts which requests
only a command's root name be displayed so in the example about we would see perl
instead if /usr/bin/per.
The following 3 successive displays are the result of
monitoring a processes named thread.pl which creates a couple of threads 10
seconds apart which then do some I/O. In the first display we see the main script, which is
actually run under the perl interpretter and so the string thread does exist as part of
the argument string, but I chose to leave it off the output to save screen real estate:
collectl --top io --procfilt fthread --procopt t
# PROCESS SUMMARY (faults are /sec) 06:57:42
# PID User PR PPID S VSZ RSS CP SysT UsrT Pct AccuTime RKB WKB MajF MinF CommandnF Command
7024 root 20 6725 S 61M 2M 2 0.00 0.00 0 0:00.00 0 0 0 0 /usr/bin/perl
A few seconds later the first thread starts up and immediately goes to the top of the list since
it does have the dominant I/O:
# PROCESS SUMMARY (faults are /sec) 06:57:52
# PID User PR PPID S VSZ RSS CP SysT UsrT Pct AccuTime RKB WKB MajF MinF Command
7065+ root 20 6725 R 73M 5M 0 0.88 0.12 100 0:01.98 0 291K 0 0 thread.pl
7064 root 20 6725 S 73M 5M 2 0.88 0.11 99 0:01.98 0 0 0 0 /usr/bin/perl
And still later the second thread shows up, it too having a higher sort order than the root script:
Naturally, as with all other data in collectl, you can record it to a file and play it
back later using various combinations of options with --procopts and even
--top. If you are using --top collectl simply displays a blank line
between intervals. Also don't forget to try different sort options
and experioment with the number of lines per process interval you want to
examine since the default is your screen height and may be too big for
playback purposes.
Including non-process data
The native top can natively show other types of data besides the top processes
and so can collectl. Just specify those subsystems you are interested in with -s and
they will be displayed in a scrolling window above the process data - by
scrolling multiple lines of data, you are able to see history, something the
linux command cannot do. You may also want to include timestamps with the
brief data by using -oT to make it easier to read.
But don't stop with brief data, you can even show verbose data as well. However
in the case of multiple subsystems it just isn't practical to show scrolling
history and so you will only see the latest sample. If you choose to show
a single verbose subsystem you will see scrolled data.
Finally, if you want to customize the way the screen real-estate is allocated between
the process and other data, you can change the size of the process section
by including the number of lines to display as the second argument to
--top. You can also control the size of the subsystem data with
--hr lines, a synonym for --headerrpeat lines.
Experimental --export proctree
This is an experimental (meaning subject to change) alternate format for displaying
process data. Rather than simply show processes in the defaut PID order or sorted
by a particular field when using the --top format, this format displays processes
in a parent/child relationship. As with all --export formats, one can use
this interactively, when playing
back data or to send the data over a socket when using --address.
At the very least, this could offer a good
starting point for developing your own alternative process output formats.
There are actually 2 main functional components to this format, the main one
being to determine the parent child relationship between all processes (there IS
some additional overhead involved here). A second function is the aggregation of
various counters and meters.
Proctree can also be combined with --top to limit the number of processes display OR
in playback mode with or without --top. Consider the following output when playing
back a file with --top --export proctree:
One can quickly see that the total CPU consumption for this monitoring interval is 0.03 of
both system and user time by simply looking at root process 1. Furthermore, of the system time
0.01 is consumed by 05535 while the other 0.02 is consumed by one of the children of 05452, actually
the grandchild 05474. One should also note that any processes with no CPU time will be excluded
from the display to keep the output reasonable dense. Without --top all processes are shown.
One can also use most of the process options as well (see --showsubopts for the complete set).
Additional interactive --top options
Proctree was really developed for real-time display with --top and so there are more available
options, the main one to consider is the suppression of fields with zero in them. In the previous
example, fields with 0 CPU were suppressed because by default --top sorts by CPU (even though
we're not sorting). If one were to choose a different sort field with --top, proctree will
use that field to suppress entries with zero in them. In fact, there are a number
of different switches one can select interactively, one of which is to change the suppression
value from 0 to something else.
So let's take a closer look at running in interactive mode by typing the command
collectl --top --export proctree
and at some time after the first data screen is displayed, type RETURN. You will now see
a menu like this:
Enter a command and RETURN while in display mode:
pid only display this pid and its children
a toggle aggregation between 'on' and 'off'
dxx change display hierarchy depth to xx
i change display format to 'I/O'
k toggle multiplication of I/O numbers by 1024 between 'on' and 'off'
m change display format to 'memory'
p change display format to 'process'
h show this menu
stype where 'type' is a valid sorting type (see --showtopopts)
entries with 0s in those field(s) will be skipped
wxx max width for display of command arguments
z toggle 'skip' logic between 'on' and 'off'
Zxx when skipping, only keep entries with I/O fields > xxKB
Press RETURN to go back to display mode...
This list shows a number of commands which will change the display contents and/or
format much in the way you can do with the standard linux top command. First type
RETURN to get back into real-time display mode and then simply type
a command at any time while collectl is running and the command will take effect on the
next display cycle.
These commands fall into several categories, one being those that toggle behavior
such as aggregation, multiplication and the skip logic. By default, all values are
aggregated up through their parent hierarchy and typing the a command followed by a
RETURN will turn this
behavior off. Similarly, when the values of the I/O counters are too large to easily read
you can force their division by 1K with the k command. And finally, you can disable the
logic that skips zero-based entries with the z command. If you'd rather skip on some
value other 0 you can set the skip value with Zxxx.
Look at the display line above the following process data:
Process Tree 09:06:03 [skip when 'time'<=0 is 'on' aggr: 'on' x1024: 'off' depth 5]
# PID PPID User PR S VSZ RSS CP SysT UsrT Pct AccuTime MajF MinF Command
00001 0 root 15 S 674M 272M 1 0.00 0.06 6 0:09.96 0 0 init
01766 1 root 15 S 50M 24M 0 0.00 0.06 6 0:01.30 0 0 /usr/sbin/sshd
02142 1766 root 15 S 25M 14M 1 0.00 0.06 6 0:00.88 0 0 /usr/sbin/sshd
02144 2142 root 15 S 18M 12M 1 0.00 0.06 6 0:00.87 0 0 -bash
02229 2144 root 19 R 14M 10M 0 0.00 0.06 6 0:00.84 0 0 /usr/bin/perl
Following the time field you can see what the toggle states are of the three fields as well
as the skip value and display depth. By default you only see 5 levels of the process
hierarchy but can change this with the d switch. For example d7 will set the depth to 7.
As with other process displays, you can also choose whether you want to see the default display, one
that shows all I/O fields or one focused on memory using the p, i or m commands. You can easily
switch between these formats at any time.
If you enter a number as a command, this is interpretted as
a process PID and the display will be adjusted such that this becomes the first entry in the
display. If you would like to skip on something other than the current field, you can easily
change that with the s command immediately following by one of the sort field names listed with
--showtop. Finally, if using the wide command option with --procopts w, long command
string will cause wrapping and make the display unreadable. The w command can be used to set
the maximum width of the command field.
As with other collectl options, there are simply far too many combinations to describe which
are appropriate for a particular situation (such as using --procopt) so it is recommended
you experiment to better understand the many capabilities of proctree.
Process Analysis
If you've run collectl as a daemon and collected process data, you now have a huge pile of
data and it's not entirely clear what you could do with. However, when that data is
played back with collectl --procanalyze, it generates a process summary file
with the extension of prcs in the same direcory as specified with the -f switch.
This file will contain one line for each unique process and the fields will be separated
by collectl's field separator which by default is a space but something you can also change
with the --sep switch.
The fields themselves summarize all the key data elements associated with each process making
it possible to see the process start/end times, cpu consumption, I/O (if the kernel supports
I/O stats), page faults and even the ranges of the different types of memory consumed. And since
the data elements are separated by a single character delimeter you can easily load the file
into your favorite spreadsheet and perform deeper analysis (the data is actually not very user
friendly as written).
It is also important to remember a couple of things:
Not all processes show up in the output if they were created and exited between a single pair of samples
You can control the time frame that is analyzed by using the --from and
-thru switches which can be a
useful thing to do when you're interesed in a specific time period and don't want this file
too cluttered
The --procanalyze overrides collectl's default behavior of processing all the data
that has been collected and so will only generate the prcs file(s). If you want to
generate plot data for other subsystems at the same time be sure to include then with -s.
By default, all data is written in space-separated format which is collectl's standard
default. Since command arguments (if any) are also space-separated they will each show up
in a different spreadsheet cell, which shouldn't be a big deal but if you want the entire
command string together, you can always use --sep 9 to make the data tab-delimited
or even choose a different separator.
Understanding Processing Overhead
This is intended to be a brief description of how process monitoring works with
the hope that it will help one use the capability more efficiently and avoid unnecessary
processing overhead. Normally the overhead is modest, but if you intend to run at higher
monitoring rates or looking at threads it's worth reading further.
Collectl maintains 2 data structures that control monitoring:
pids-to-monitor and pids-to-ignore.
These lists are built at the time collectl starts, so if
--procopts p is not specified, the effect is to execute a ps command
and save all the
pids in the pids-to-monitor list. If filters are specified with
--procfilt, only those
pids that match are placed in pids-to-monitor list and the rest placed in the
pids-to-ignore list and so you can see that when filters are used there can be
a significant reduction in overhead since collectl need not examine every
processes data.
If collectl is only monitoring a specific set of processes, either because --procopts p
was specified or procfilt was used and only specified specific pids (not ppids), on
each monitoring pass collectl only looks at the pids in the to-be-monitored
list. In other words, this is as efficient as it gets because it needn't look
for processes if neither list, aka newly created processes.
If doing dynamic process monitoring, every monitoring pass collectl has to
read all the pids in /proc to get a list of ALL current processes. While it ignores any in
the do-not-monitor list, it must look at the rest. If any of these are in the
to-be-monitored list and have had thread monitoring requested, additional work
is required to see if any new threads have shown up.
Any processes not in the to-be-monitored list are obviously NEW processes and must
then be examined to see if they
match any selection criteria and this involves
reading the /proc/pid/stat file. That pid is
then placed in one of the two lists.
It should be understood that during any particular interval a lot of processes
come and go, such as cat, ls, etc. However, these are short lived enough as to
not even be seen by collectl, unless of course collectl is running at a very
fine grained monitoring level.
Occasionally a process being monitored disappears because it had terminated.
When this happens its pid is removed from the to-be-monitored list.
Finally, these data structures (and a couple of others that have not been
described) need maintenance to keep them from growing. If the number of
processes to monitor has been fixed, this maintenance is significantly reduced.
So the bottom line is if you have to use dynamic monitoring, try to bound the
number of processes and/or threads. If you really need to see it all, don't
be afraid to but just be mindful of the overhead. Collecting all process
data with the default interval has been observed to take about 1 minute of CPU
time, which is less than 0.1%, on a lightly loaded Proliant DL380, but that load
will be higher with more active process.
RESTRICTIONS
You cannot specify --procfilt during playback mode. If you need to look at a subset of
the data consider using a filter like grep.
Thread monitoring is limited to 2.6 kernels.
Process I/O monitoring is limited to kernels that have that capability enabled and that
didn't even appear before 2.6.22. If you don't see the file /proc/self/io, your kernel
was not built with process I/O accounting enabled and you need to get one that has the following
enabled: CONFIG_TASKSTATS, CONFIG_TASK_XACCT and CONFIG_TASK_IO_ACCOUNTING.
There is a bug in the way the kernel reports I/O stats (see bugzilla
10702) such
that when you exclude threads from the display the parent process I/O counts are not
aggregated into the I/O counts from the threads resulting in misleading results.
Andrea Righi has provided a patch here that
provides the correct aggregation, but of course that will require a kernel rebuild. He
has also informed me that his patch has been included in the 2.6.26 kernel so this should no
longer be an issue from that version forward.
Most modern computer systems have the capability of reporting temperatures and fan
speeds (as well as other information) via ipmi. However, the format of the device
names is not standardized making it extremely difficult if not impossible to
programmatically interpret and report it. This feature has been declared to be
experimental in order to evaluate its success on a broader set of hardware,
which I expect will be determined by the number of bug reports. As such it is
disabled in collectl.conf so if you want to enable it when running as a daemon
be sure to include an E in the -s string in the DeamonCommands line as
shown below:
Collectl depends on the open source tool ipmitool,
which must be installed first. Installation is as simple as pulling down and unpacking the kit
with tar and executing the commands:
./configure
make
make install
That's all it takes. However, collectl must be run as root and your system must
support ipmi. The easiest way to tell is
if the command dmidecode | grep IPMI runs without error. If you get the error
Could not open device at /dev/ipmi0... you are not running as root.
If you get some other error your system probably
does not support ipmi and even if you were able to install impitool, you won't be able
to use it.
The next step is to start the ipmi driver, and this is generally done via the command
service ipmi start on a RedHat system or something line /etc/init.d/impi start
on others. On some systems such as HP blades, you may need to install a
custom ipmi driver such as hp-OpenIPMI and start that instead of the standard
driver.
At this point you should be able to execute the command "ipmitool sdr and see a all
your sensor data or the commands ipmitool sdr type fan and ipmitool sdr type
temp to just see fan and temperature data:
[root@bl460-63 ipmitool-1.8.9]# ipmitool sdr
UID Light | 0 unspecified | ok
Int. Health LED | 0 unspecified | ok
VRM 1 | 0 unspecified | cr
VRM 2 | 0 unspecified | cr
Temp 1 | 47 degrees C | ok
Temp 2 | 34 degrees C | ok
Temp 3 | 30 degrees C | ok
Temp 4 | 30 degrees C | ok
Temp 5 | 31 degrees C | ok
Temp 6 | 30 degrees C | ok
Temp 7 | 30 degrees C | ok
Temp 8 | 66 degrees C | ok
Temp 9 | 20 degrees C | ok
Virtual Fan | 37.24 unspecifi | nc
Enclosure Status | 0 unspecified | nc
The collectl interface
Collectl uses a 3rd monitoring interval to collect ipmi data, which by default is 2 minutes.
This is done for several reasons:
The ipmitool interface involves running another process and adds to the load,
using a less frequent interval helps minimize collectl's footprint
This type of data changes at a slow enough rate as to not require monitoring too frequently
although the current readings which have just been added to Version 3.1.2 do seem to
change in nead real-time and so the monitoring interval may need to be rethought
Although you can run collectl to report this data interatively to report a single sample,
its typical use is expected to be as a daemon. When run as a daemon collectl and -sE
is specified, which is currently the default, it will first check to see if
ipmitool is present and if a communications device of the form /dev/ipmi*
is present. If so it will start collecting ipmi data for fans and temperature sensors.
If it can detect a current sensor is present, it will report on that too.
You can control the way ipmi data is displayed in playback mode using --envopts
and one of 3 switches that allow you to only report fan or temperature data and if you
are reporting both, which is the default, you can request the 2 types of data be displayed
on separate lines. This latter option can be useful if you have a lot of devices on
which to report.
The following is an example of time-stamped output on an HP BL460c Blade, first without any options
Here we see the effect of --envopts M when examining an HP DL380-G5.
However it does also generate a lot more
noise in the output. It's main purpose is for dealing
with too much data to comfortably display on a single line.
If you choose to convert the data to plot format, a file with the extension env
will be created.
Device Names and the collectl challenge
As already mentioned, there is no standard on how one names an ipmi device and as a
result the names used even on the small sample of systems tested during development
have been quite different. Here is are just a few ways fan names are reported:
Fan 1
Fans
CPU FAN1
SYS FAN1
Fan1A (CPU)
FAN CPU0
FAN MOD 1A RPM
Fan Redundancy
On the one hand, collectl could simply report the exact names as they are reported,
but the challenge of trying to format them in such a way as to provide a compact
display are impossible. Given that the collectl standard reporting format is a
single data header line, the notion of multiple-line headers is not an option.
While it is tempting to simply determine the widest device name and use that for
a header width, for systems that report over a dozen devices you couldn't fit them
on the same line and that's only for systems that have been tested.
After looking at all these different names and formats, one common theme did emerge.
All devices appear to have optional numbers (I didn't see any with just letters)
and those numbers if there have optional letters. Furthermore, there seems to be some sort of
optional type associated with many as well. This led to the idea of a
standard naming for these devices as follows:
[type]Fan|Temp[devicenumber[deviceletter]]
in which the type field would be limited to a single character. Applying this
scheme to the examples above leads to the following name mapping:
Fan 1 Fan1
Fans Fan
CPU FAN1 CFAN1
SYS FAN1 SFAN1
Fan1A (CPU) CFan1A
FAN CPU0 CFAN0
FAN MOD 1A RPM MFAN1A
Fan Redundancy RFan
This is admittedly not perfect but seems like a reasonable compromise and since collectl
will report the device names in the same order returned by ipmitool it is not all that
difficult to figure out how collectl chose to map them.
Parsing Names and Customization
This is where the fun begins or things get really ugly, depending on your perspective.
After examing many different types of device name formats, it was determined that most
tended to follow the pattern of
prefix type instanceNumber suffix
Where things get a little crazy is that sometimes the actual instance number can be part
of the prefix OR sometimes the instance contains a letter.
All that said, collectl breaks a device name into these components, assuming a numeric
instance. It then applies the minimal set of tests/modifications, note there are examples
of all these cases in the sample names shown earlier:
if the suffix contains the string in ()s, set the prefix to the string contained within
if an instance and suffix, append first word of suffix to the instance
if no prefix or instance and the suffix doesn't start with a digit, set the prefix to the suffix
if there is a prefix, prepend the first letter to the device name which is fan or temp
if there is a prefix and it contains a digit, use that as the instance number
remove all whitespace
Since this can get verf confusing, a special switch names --envdebug has been included
which will show the actual parsing of the device names. The following is an example of parsing
some of the names listed above:
Fan CPU0 Tach,3480
Prefix: Name: Fan Instance: Suffix: CPU0 Tach
Fan1A (CPU),EAh,ok,29.3,Performance Met
Prefix: Name: Fan Instance: 1 Suffix: A (CPU)
FAN MOD 1A RPM,5775,RPM,ok
Prefix: Name: FAN Instance: Suffix: MOD 1A RPM
In the first example, the prefix is set to CPU0. Then in instance set to 0 and the
C prepended to Fan
In the second case the prefix is set to CPU and ultimately the name resolved to CFan1A
In the third case the prefix is set to MOD and the device name set to MFAN and the
instance number is lost when what we really want is to have it set to 1A!!!
Standard and User Defined Parsing Rules
Collectl provides a mechanism for dealing with device names that do not result in the
generation of satisfactory names as described in the last section, by providing a file
of translation rules for the system(s) they apply to.
This file is currently populated with a number of HP systems this mechanism has
been tested on and the rules are actually one or more perl pattern matching/replacement
expressions. If the system name can be obtained via dmidecode, this file
is searched for a matching stanza and its translations applied to device names
before their initial parsing and/or immediately after a device name is generated.
If your system is not in this standard set, you can either add your own rules to
/usr/share/collectl/envrules.std (assuming your system type can be obained
through dmidecode) or put them in a standalone file and tell collectl to use it
instead of the standard one using --envrules. If you do use your own file it
should simply contains line of the following form (no stanza preface) noting that spaces
and comments (lines preceeded with a #) are permitted:
If you know perl (and you really should if you want to do this), collectl builds a perl
pattern match command if you specify [ignore] and ignores any strings returned by
ipmitool that match. This is a good way to reduce the volume of sensors on systems that
may have dozens of them and you're only interested in a specific subset.
In the cases of [pre] and [post] a perl substistituion command is built out
of the pattern and replace strings and applied to the sensor names. There
is one caveat about [post] and that is it only applies to the actual derived sensor
name and not the instance, so it you want to change a specific instance consider
using a pre string to make a unique sensor name and then change it to what you
really want with post. So looking at the string
FAN MOD 1A RPM
and the processing rules described in the previous section, the MOD
suffix will be prepended to FAN and the first letter used to name the device
MFAN, losing the instance information with is 1A.
There are at least 3 options here. The first is to simply remove MOD from
each name which we can do with the rule:
/ MOD//
which will result in
the instance names being picked up correctly because they will now immediately follow
FAN. In fact, if you include --envdebug along with your rules you'll
see the results of the replacement:
FAN MOD 1A RPM,5775,RPM,ok
Pre-Remapped 'FAN MOD 1A RPM' to 'FAN 1A RPM'
Prefix: Name: FAN Instance: 1 Suffix: A RPM
The second option would be to move
MOD to the front of the string so that the rule that uses the first letter
as part of the final name will take effect and that rule will look like:
/(.*) MOD (.*)/MOD $1$2/
and results in the following parsing:
FAN MOD 1A RPM,5775,RPM,ok
Pre-Remapped 'FAN MOD 1A RPM' to 'MOD FAN 1A RPM'
Prefix: MOD Name: FAN Instance: 1 Suffix: A RPM
Unfortunately in order to make perl iterpret the $1$2 symbols an eval
is required which generates a little extra overhead and while not horrible an
even better solution is the third option which doesn't use any special $ symbols:
/FAN MOD/MOD FAN/
which produces exactly the same results as the previous example except without the
eval command.
There is in fact at least one other mechanism for those that are not all that familiar
with perl and is only being included for completeness, and that is to simply hardcode
the replacement of each device with the desired output. In other words
/FAN MOD 1A RPM/MOD FAN1 A/
/FAN MOD 2A RPM/MOD FAN2 A/
/FAN MOD 3A RPM/MOD FAN3 A/
etc
will produce strings that can also be properly parsed without involved $ variables
but this means you need to specify each unique device name to remap and it will also result
in all pattern matching statements to be executed for each device which will also
result in slightly more overhead.
Power Monitoring
Currently all the systems that power monitoring has been testing on report it as the field
Power Meter and without more examples, the parsing is currently set up to specifically
look for that field.
Performance and Alternate IMPI Devices
In some situations there may be multiple ipmi devices over which to communicate and if so, the
default one may not necessarily be the fastest one. If you thing the ipmi commands are taking
too long to execute, try a simple experiment like this:
ipmitool sdr dump /tmp/xxx
time for i in `seq 1 10`; do ipmitool -S /tmp/xxx sdr > /dev/null; done;
real 0m20.476s
user 0m0.004s
sys 0m0.015s
As you can see, even though the command only used 0.02 seconds of CPU time, the elapsed time
was over 20 seconds, a good indication something is not right. If look in /proc/ipmi you make
see more than one directory as in the following case:
[root@hpdc3dmgt1 ~]# ls /proc/ipmi
0 1
This means there are 2 different IPMI devices and since the default is one, let's try repating
the command above on the other device. Also notice that since we've already initialized
our cache file we do not need to reissue the ipmitool sdr dump command:
time for i in `seq 1 10`; do ipmitool -S /tmp/xxx -d1 sdr > /dev/null; done;
real 0m0.487s
user 0m0.004s
sys 0m0.013s
See how the elapsed time is only a fraction using device 1? To tell collectl to use this
device instead of the default, simply specify the number in the --envopts switch,
for example collectl -sE --envopts 1
Restrictions
Some systems report what appears to be device codes in the data field and the data in the 4th
field and I don't know why. For now, when this occurs report the 4th column as the data
instead. If this breaks other things it will have to be removed and invalid data reported
for those who do not report it in column 2.
Not really a whole lot to say here other than collectl does not report any socket details at this time but only
summary data, which it gets from /proc/net/sockstat. In brief format the data reported is the number of
sockets in use by their type, specifically TCP, UDP and Raw. It also reports the number if IP fragments.
In verbose more it break things out at lower levels of detail.
In most cases this is of minimal interest unless you're trying to track down a specific socket related problem. In the cases
of a runaway process or someone opening but not closing sockets this number has been seen to grow quite large and even
consume all resources causing a system crash, but those cases are pretty rare. In any event, during times of strange behavior
it can't hurt to have a look at these numbers if for no other reason than to rule out socket problems.
As with --import, the --export option allows one to build
custom modules, but in this case for generating output. Unlike --import, which
allows multiple modules to be specified, you only --export
using a single module, which in turn overrides all of collectl's output
formatting routines.
These modules all end in the extension .ph and are searched for first
in the directory collectl is being executed from and then /usr/share/collectl.
If the module name is prepended with a directory name, collectl will search only
there.
The reason you might care about this is that now if you want to produce your
own exportable form of output and be able to print it locally, make it
available to another program over a socket or even write to a local file
while still being able to log to raw and/or plot formats,
you get that all that functionality for free.
How It Works
The interface to all this is really quite simple. At the command line the user
types collectl --export name[,options] where:
name both names a file to be required by collectl as well as
the name of the entry points to both an initialization routine as well as one
that produces the output
options specifies an optional list of arguments that are passed to both
routines to do with whatever they wish
There are currently 6 different custom exports that are part of a standard collectl
distribution:
sexpr generates output as an s-expressionsnote: sexp is deprecated and will be removed from the kit in 2014
proctree provides another way to look at process data
vmstat is the standard unix vmstat command most people know and love,
so why reinvent it? It's both a useful example of how to write an export as well as
provides the benefit of allowing one to play back collectl data in vmstat format!
The first four are the most interesting because they have been built to take advantage
of collectl's capability of sending their output over a socket and are therefore the ideal
vehicle for interfacing with other tools and environments. All four share a common set
of options, though sexpr is fairly limited, as described in the following table:
NOTE: sexpr is deprecated and will be remove from kit in 2014
align
Used in conjunction with i= and when specified data samples will
be aligned to whole minute boundaries. In other words if used with i=15, data will be reported
at the top of the minute, 15 seconds past, etc. The first sample may therefore be partial.
avg|max|min|tot
used in conjunction with i=, send the average, maximum,
minimum or total of the data over the associated set of intervals specified by i=. If
none of these are specified, the values from the most recent monitoring interval will
be reported.
co
this does not take a value and indicates changes only such
that only data elements that have changed since they were last sampled are reported in
an attempt to minimize processing and network bandwidth. If not specified, samples for
all reporting intervals will be sent.
d=mask
debugging mask, see beginning of the actual export file for details
f=file
names the output snapshot file, which applies to only lexpr and sexpr.
If this option is not used, -f must be and the snapshot file names are then set to the
single character L and S respectively and written into the directory
associated with -f
caution
If you are writing your own export module that doesn't use a snapshot file, you must explicity
include error checking to assure it is not run without at least -P or --rawtoo in
combination with -f
h
shows help/usage
i=secs
specifies the reporting interval in seconds. In other words, if you specify
i=60, a sample will be reported every 60 seconds independent of collectl's monitoring
interval. The default is to report every sample. This interval must be a multiple of
the base collectl interval. note that while collectl always rounds rates to the next whole
value, when multiple intervals are added together only the totals are rounded.
s
specifies a subset of those subsystems specified with -s in the collectl command line
and only data collected for that subset will be reported. The default to report everything.
In the case where you only want report data collected via --import, use s= with
no args.
ttl
is the time to live in intervals for each piece of performance data. If
more than this number of intervals passes data will be sent regardless of whether it changed
or not and the ttl countdown timer reset. The default is 5. This actually has a second use
for gexpr and that is to set the gmond ttl to double this number multiplied by the interval.
Logging
Collectl can actually create up to 3 different type of log files and it's worth spending a little more
time enumerating how collectl decides where and when to create them.
In its simplest form, when one runs collectl with --export and no options, all output is sent to the
terminal, noting gexpr and graphite always require an address.
If collectl is run with -f, and one is exporting lexpr or sexpr a snapshot
file is created in the directory associated with -f.
If a socket is to be used by specifying collectl -A and the export supports
socket I/O:
all output goes out over the socket and no snapshot file is involved
if one specifies -f, a raw file is created in the directory specified by -f.
If -P is also used, a plot file is created instead and if -P and --rawtoo are used,
both types of files are created. caution: this is different behavior than one sees when running
without sockets in which case a snapshot file is written to the directory named by -f
Since gexpr and graphite do not use -A for specifying socket I/O and do their own
communications, collectl's rules for non-socket logging apply, in which case -f by itself implies
a snapshot file is involved and so requires -P or --rawtoo or both to do any logging.
To create the snapshot file in a directory other then the one specified with -f, use the
f= option with --export
If one tries an invalid combination of switches or something not supported, it is up to
the export itself to determine that since by design collectl has no knowledge of those
module's inner workings.
Example
Perhaps the best way to see how all this works is with a simple example and
it turns out that vmstat.ph is small enough to meet that need. You
may also wish to refer to the others as well to see how some of the more exotic
capabilities are implemented.
This first section gets called almost immediately by collectl after reading
in the various user switches. This is the place to catch switch errors
and since this routine always requires -scm we'll just hardcode it
to that and reject any user entered ones. This initialization subroutine
must be named for our module followed by Init.
sub vmstatInit
{
error("-s not allowed with 'vmstat'") if $userSubsys ne '';
error("-f requires either --rawtoo or -P") if $filename ne '' && !$rawtooFlag && !$plotFlag;
error("-P or --rawtoo require -f") if $filename eq '' && ($rawtooFlag || $plotFlag);
$subsys=$userSubsys='cm';
}
Next we define the output routine, with the same base name as that
of our included file.
The if statement uses collectl's standard idiom for printing headers
based on the number of lines printed and whether or not the user wants only a
single header, no header or even to clear the screen between headers. If you
do not want/need all these features it's perfectly fine to use a more simplified
header printing mechanism, such as the one is sexpr.ph.
sub vmstat
{
my $line;
if (printHeader())
{
$line= "${cls}#${miniBlanks}procs ---------------memory (KB)--------------- --swaps-- -----io---- --system-- ----cpu-----\n";
$line.="#$miniDateTime r b swpd free buff cache inact active si so bi bo in cs us sy id wa\n";
}
Next comes the handling of optional date/time prefixes that I stole from
printTerm() in formatit.ph and which can be controlled
by various switch options. Again, if you have no intent of supporting these
you can even put in error handling in your initialization routine
or simply ignore the switches.
my $datetime='';
if ($options=~/[dDTm]/)
{
($ss, $mm, $hh, $mday, $mon, $year)=localtime($lastSecs);
$datetime=sprintf("%02d:%02d:%02d", $hh, $mm, $ss);
$datetime=sprintf("%02d/%02d %s", $mon+1, $mday, $datetime) if $options=~/d/;
$datetime=sprintf("%04d%02d%02d %s", $year+1900, $mon+1, $mday, $datetime) if $options=~/D/;
$datetime.=".$usecs" if ($options=~/m/);
$datetime.=" ";
}
Here we build the actual output, noting that we're not really printing
anything yet, but rather building up a string (which may contain the header)
that we will print in one shot.
Finally comes the output. There is actually a lot of latitude here and in this case
we're calling printText() which will send the output to the terminal or over a
socket. It will not write to a local file as do sexpr or lexpr, but if you
want to see how to do that, refer to them. As with all perl require files, they must
return true and therefore the final line is the digit 1.
printText($line);
}
1;
Try running it and you'll see all the pagination and time formats work just as they
do with standard output formats.
All files generated by collectl via the -f switch, both raw and plot, will
always contain the name of the host from which they have been generated
according to the following rules:
If the specified name is a directory, the resultant file(s) will be
created in that directory and begin with the hostname. If the name is not a
directory, that name will be prepended to -hostname
The name is then followed with -yyyymmdd
If this is a raw data file or one generated using -P and -ou
has been specified, it will also have -hhmmss appended as well to
indicate the starting time of the sample. Note that the colons have been left off the
time field to make it easy to move the file to a PC for further analysis if
so desired.
The appropriate extension is added and if a compressed file, .gz
is then appended.
warning
Never ever try to rename a file and expect collectl to be able to process it in
playback mode. When running in this mode, collectl ignores any files that do not
look like they were generated by collectl. It also verifies what appears in the hostname
portion of the filename to match that which is recorded in the header.
The first thing to understand about lustre reporting is in most cases, where one has configured
the server(s) and just wants to monitor them, all one need do is specify -sl or -sL and
collectl will do the right thing.
It will automatically detect the type of service(s) currently running and will
either record or display the appropriate data.
If you select -sl and the system doesn't have lustre installed, it will warn
you and then disable that switch.
Controlling Which Data is Displayed
Lustre records a wealth of performance data, far more than makes sense to
display all the time, and so by default collectl displays minimal information
such as bytes/operations read and written. At the client detail level lustre
can differentiate this
data at the filesystem and even the OST level! In order to accommodate the broadest flexibility
one is allowed to control the way data is collected/displayed via several complementary
switches.
-s: As is normally the case, one can specify '-sl' for summary level data, '-sL' for
detail data or combine them to get both. However, since the client detail data can actually
be presented at the individual filesystem or OST level, there is an option to show the OST
level details (filesystem details are the default) see --lustopts O.
--lustopts: This switch is used to provide further detail about the types of data that
is to be collected/displayed. There are 5 such values that collectl cares about:
B - rpc buffer level data.
D - disk block statistics, which applies to both MDS and OSS servers. One should also note
this is specific to HP SFS and this data is not available in the open source version.
M - client metadata (note that this was the default prior to collectl V1.6.2).
O - for client details only, show results by OST
R - read_ahead statistics. Unlike the other options, which generate a lot of data,
--lustopts R may be used with brief mode.
As it turns out, nothing is quite as simple as it seems and while the following
case is not typical, it needs to be addressed for completeness. Since collectl allows one to
collect one set of data and to later display a different set,
consider what happens in one were
to collect multiple types of lustre data for a client using --lustopts MR, but then just play back
the basic client data which is collected without specifying --lustopts.
By default, playback mode defaults
to the settings data was collected with and to change the display one needs to explicitly
change those settings. To meet this need, use --lustsvc, which is described in more
detail later.
In the spirit of letting the user display whatever they want to, collectl will allow one to
select multiple values for --lustopts and it will try to display the results appropriately.
Perhaps the easiest thing to do is just experiment and in most cases you'll get what you're
looking for. There are a few combinations of -s and --lustopts that do not
make sense and if you choose one, you will be told.
What About Playback?
As is always the case with playback, unless otherwise told to do something else, collectl
will playback its recorded data
based on the parameters selected for collection. In other words, if you specify
--lustopts OBR
in record mode, collectl will record both RPC buffer and read_ahead stats. When you play the
data back, it will then display both as well. However, you also have the option of specifying
--lustopts, typically thought of as a collection-only switch, and it will force the output to what
you'd like it to be. If you select a statistics type that hasn't been recorded,
that information will be displayed, but as zeros.
Recognizing Service Configuration Changes
In some cases lustre services may change after collectl starts. In fact, it may not
even be running and if so you'll get a message telling you it is not and that collectl
cannot determine the system type since it could be a client, MDS, OSS or some
combination. This includes services starting and stopping as well as the configurations
of those services themselves changing.
For example one might occasionally mount/umount different lustre filesystems on
a client. Not to worry. Collectl
periodically checks for configuration changes and automatically adjusts the data it collects
as well as anything it may be currently displaying. However this can also lead to the
output format changing. If you know that the system type could change and you simply want to
force the type of output to be consistent, use --lustsvc as described
in the next section.
Changing the Default Recording/Display Behavior
There are some times when you want specific control over what data is recorded or
displayed rather than the default behavior OR collectl starts before lustre does and
it can't determine the type of system it is. This is typically the case when a system
is playing multiple roles by providing more than one service. For example, if a system
has been configured as both an OSS and a client, every time you run collectl you will collect
or display data about both and sometimes this is NOT what you want. There may be other times
where you have developed some reports or graphs that expect data in a standard format and
you've collected a subset (or superset) of data.
To override this behavior of the lustre portion of the data (remember you can
control the displaying of individual subsystems with -s), use --lustsvc to
specify the type of service(s) you're interested
in and collectl will only pay attention to those, both for recording to a file as well as
display. Naturally when displaying
data for services you never collectled data on, those services will print as zeros.
If all this sounds confusing, just experiment with various combinations of -s,
--lustopts and --lustsvcs and observe the behavior.
The mechanism for including custom recording/reporting code into collectl is very similar to that for
exporting custom data. One uses the switch --import followed by one or more file names, separated
by colons. Following each file name are one or more file-specific arguments which if specified are
comma separated as shown below:
collectl --import file1,d:file2
In this example collectl will look for the files file1.ph and file2.ph, noting
that the first has the single argument 'd'. Collectl will execute a perl require on each
file (in the order they're specified) and subsequently call functions in them from various locations
within collectl. Looking for strings in both collectl and formatit.ph that begin
with &{$imp will identify the locations where collectl calls the functions named in the API
and may help during the development/testing process to better understand what collectl is doing.
As a reference, a simple module has been included in the same main directory as collectl itself,
which is named hello.ph as collectl's version of Hello World. Since it can't read
anything from /proc it is hardcoded to generate 3 lines of data with increasing data values. Beyond
that bit a hand-waving, everything else it does is fully functional. You can mix its output with
any standard collectl data, record to raw or plot files, play back the data and even send its output
over a socket.
From time to time additional import modules may be included in collectl which may also be used
as reference. For example, the module misc.ph is now also part of collectl. It imports data
about the uptime, number of people logged in, the cpu frequency and the number of mounted nfs file
systems.
It should be noted that although collectl itself does not use strict, which is a long story,
it is recommended these routines do. This will help ensure that they do not accidentally reuse a
variable of the same name that collectl does and accidentally step on it.
A couple of words about performance
One of the key design objectives for collectl is efficiency and it is indeed very lightweight,
typically using less than 0.2% of the CPU when sampling once every 10 seconds. Another way to
look at this is it often uses less than 192 CPU seconds in the course of an entire day.
If you care about overhead, and you should, be sure to be as efficient as you can in your own code.
If you have to run a command to get your data instead of reading it from /proc, that will be more
expensive. If that command has to do a lot of work, it will be even more expensive.
It is recommended your take advantage of collectl's
built-in mechanism for
measuring its own performance. For example, measuring the performance of the hello.ph module,
which does almost nothing since it doesn't even look at /proc data, uses less than 1 second to read 8840
samples on an older 2GHz system, which is the equivalent of a full day's worth of sampling. Monitoring
CPU performance data take about 3-1/2 seconds and memory counters take about
7 seconds, just to give a few examples of the more efficient types of data it collects.
Access to collectl internal functions, variable and constants
Collectl is relatively big, at least for a perl script, consisting of over 100 internal subroutines,
most of which are simply for internal
housekeeping, but some of which are of a more general purpose. It also keeps most of its statistical
data in single variables and one dimensional arrays. Clearly hashes could make it more convenient
for passing data around but it was felt that the use of more complex data structures would generate
more overhead and so their use has been minimized.
While it is literally impossible to enumerate them all, there are a relatively small number of
functions, variables and constants that should be considered when writing your routines to
insure a more seamless integration with collectl. The following table is really only a means
to get started. If you need more details of what a function actually does or how a variable is used,
read the code.
Function
Definition
cvt()
Convert a string to a maximum number of characters, appending 'K', 'M', etc
as appropriate. Can also be instructed to divide counters by 1000 and sizes by 1024.
error()
Report error on terminal and if logging to a message file
write a type 'E' message. Then exit
fix()
When a counter turns negative, it has wrapped. This function will convert
to a positive number by adding back the size of a 32-bit word OR a user specified data width.
getexec()
Execute a command and record its output to a raw file when operating
in collectl mode, prepended with the supplied string
getproc()
Read data from /proc, prepending a string as with getexec except in
this case you can also instruct it to skip lines at the beginning or end. See the function itself
for details
record()
Only needed if not using getproc, which will call it for you. It writes
data to a raw file> record when in record mode or calls the appropriate print routines in interactive
mode.
a single line of data
Variable
Definition
$datetime
The date/time stamp associated with the current set of data, in the user
requested format, based on the use of -o. See the constant $miniFiller which is a string of
spaces of the same width.
$intSecs
The number of seconds in the current interval. This is not an integer.
Constants
Definition
$miniFiller
A string of spaces, the same number of characters as in the
$datetime variable
$rate
A text string that is set to /secs and appended to most of
the verbose format headers, indicating rates are being displayed. However, if the user specifies -on
with the collectl command to indicate non-normalized data, it is set to /int to indicate
per-interval data is being reported.
$SEP
This is the current plot format separator character, set to a space by default,
but can be changed with --sep so never hard code spaces into your plot format output.
The API
The API between collectl and user written code is actually a fixed number of callbacks. In other
words, when you tell collectl to import a piece of code, it not only uses that name to identify
the code it also uses that name as a qualifier on the name of the functions it calls.
If you load a module called mymodule, collectl will then make calls to mymoduleInit(),
mymoduleGetData() and several others as enumerated in the table below. You must include
all these function call backs in your code or prevent them from being called by restricting which
switches the user is allow to specify in the collectl command line. For example if your module
doesn't want to support plot data and you generate an error if the user specified -P (which can be
checked by examining $plotFlag in your init routine), you can safely leave off the
PrintPlot callback.
Function
Definition
Analyze
Examine performance counters and generate values for current interval
GetData
Read performance data from /proc or any other mechanism of choice
GetHeader
During playback only, supply the header for additional initialization
Init
One time initializations are performed here
InitInterval
Initializations required for each processing cycle
IntervalEnd
Optional routine, called at end of each processing cycle if defined
PrintBrief
Build output strings for brief format
PrintExport
Build output strings for formatting by gexpr, lexpr and sexpr, which are
3 standard collectl --export modules
PrintPlot
Build output string in plot format
PrintVerbose
Build output string in verbose format
UpdateHeader
Add custom line(s) to all file headers
There are also several constants that must be passed back to collectl during intialization.
See Init() for more details.
A string consisting of 's', 'd', or 'sd', indicating whether or not this module supports
summary data, detail data or both.
A unique string for any data collected by this code and ultimately
passed to collectl in record(). As a result, all data written to the raw file will be
prepended by this discriminator so that during actual collection and/or playback collectl will know
to call the analysis routine upon seeing this type of data. Therefore this string must be unique
with respect to any other data collection qualification strings. This string will also be used as an
extension for any detail plot files and so should be on the order of 3 or 4 characters long.
Analyze($type, \$data)
This function is called for each line of recorded data that begins with the qualifier string that
has been set in Init. Any lines that don't begin with that string will never be seen by
this routine. You should also be sure that string is unque enough that you aren't passed data
you don't expect.
$type is the first space-separated field from the recorded data. It is completely under
your control to decide how to process that data. In some cases there will only be a single line to
process and no further decisions will be required. In other cases there
may be multiple lines and it may be necessary to include an additional discriminator when recording the
data so that the analysis routine can figure out what to do with it. For example, try running with
-d4 and look at the data returned for -sf or -si.
$data is a reference to the string of everything that follows $type in the current
line of data being processed
GetData()
This function takes no arguments and is responsible for reading in the data to be recorded
and processed by collectl and as such you should strive to make it as efficient as possible.
If reading data from /proc, you can probably use the getproc()
function, using 0 as the first parameter for doing generic reads. If you wish to execute
a command, you can call getexec() and pass it a 3 which is its generic argument for capturing
the entire contents of whatever command is being executed.
If you want to do your own thing you can basically do anything you want, but be sure
to call record() to actually write the data to the raw file and optionally
pass it to the analysis routine later on.
In any case, each record must use the same discriminator that Analyze is expecting
so collectl can identify that data as coming from this module. You may
also want to look at the data gathering loop inside of collectl to get a better feel for
how data collection works in general.
To make sure you're collecting data correctly, run collectl with -d4 as shown below for reading
socket data, which uses the string sock as its own discriminator. The Analyze
routine then needs to look at the second field to identify how to interpret the remainder of
the data line.
This function is called when one needs to know what is stored in the header of a file being
played back - it is only called if collectl doesn't find it and therefore optional.
While it is impossible to know how a module will use GetHeader, it is often used to
retrieve instance numbers, such as how many nvidia GPUs one might have been monitored or
their type (see nvidia.ph), both of which would have been written to the header
using a call to
UpdateHeader.
$header is a reference to the header once it is read in collectl's initFormat
Since standard collectl processing is to always playback what the user requests, even if
that data hadn't even been collected, the same holds true here. If one had gotten to
this point and it is determined there is no data, the API does not contain a failure return
code as does init. Rather, one would simply end up reporting 0s for all values.
Init(\$options, \$key)
This function is called once by collectl, before any data collection begins. If there are any
one-time initializations of variables to do, this is the place to do them. For example, when processing
incrementing variables one often subtracts the previous value from the current one and this is the
ideal place to initialize their previous values to 0. Naturally that will lead to erroneous results
for the first interval, which is why collectl never includes those stats in its output. However, if
you don't initialize them to something you will get uninitialized variable warnings the first
time they're used.
$options is a reference to the option string, if any, passed immediately
after the name following--import and specifies whether the user wants to see
summary data, detail data or both. If none specified it must be set to the
default behavior desired for this module, typically 's'.
$key is a reference to a string of usually 2-4 characters which will be
prepended to each data record and used in several places to associate data records with
this module
Upon completion return 1 to indicate success, Returning a -1 will indicate failure
and result in the imported module's functional calls to be removed from collectl's call
stack and will no longer be actively called.
InitInterval()
During each data collection interval, collectl may need to reset some counters. For example, when
processing disk data, collectl adds together all the disk stats for each interval which are then
reported as summary data. At the beginning of each interval these counters must be reset to 0
and it's at that point in the processing that this routine is called.
IntervalEnd()
As described earlier, if this routine exists it is called at the end of an interval processing cycle.
This makes it possible to do any post processing that may be require before the start of the next
interval. In many cases this is not necessary.
PrintBrief($type, \$line)
The trick with brief mode is that that multiple types of data are displayed together on the same line.
That means each imported module must append its own unique data to the current line of output as it
is being constructed without any carriage returns. Further, since there are 2 header lines and
brief format supports the ability to print running totals when one enters return during
processing, there are a number of places one needs to have their code called from.
$type is an integer from 1 to 6 and is used to decide which subfunction to execute.
Think of it this a case statement control variable.
$line is a reference to the current output line being constructed. Simply append
whatever makes sense for that function, be it a header or a value. In the case of the running
totals, since that is always printed on the terminal and never goes to a socket a simple
print is sufficient.
What about custom export modules and how this effects them?
The good news is that at least for the standard 3, sexpr, lexpr and the newest
gexpr all support --import. In other words they too have callbacks that you must respond
to if your code is being run at the same time as one of these.
Again, see hello.ph for an example, but suffice it to say you need to do something when
called, even if only a null function is supplied.
Since both lexpr and sexpr can write their output to the terminal, the easiest way
to test these is to just run collect and have it display on the terminal. However, the output
of gexpr is binary and so the easiest way to test this code is to tell it not to open a socket
(though you must supply an address/port, even if invalid) and report the 3 data elements it is about to
encode by running with a debug value of 9 noting this is gexpr's own internal debug
switch and not collectl's. The 8 bit tells gexpr to not open the sockeT (which is why an
invalid value is ok) and the 1 bit tells it to display its output nicely formatting on the terminal.
$type is a 'g', 'l' or 's' depending which of the 3 expr modules is exported and the
interpretation of the 4 references that follow are type dependent
type=g
$ref1 is the data label
$ref2 is the units string
$ref3 is data value
type=l
$ref1 is the summary data label
$ref2 is the summary data value
$ref3 is the detail data label
$ref4 is the detail data value
type=s
$ref1 is the summary data line
$ref1 is the detail data line
PrintPlot($type, \$line)
This type of output is formatted for plotting, which can get quite complicated
based on whether you are writing to a terminal, multiple files or a socket. Fortunately
all that headache is handled for you by collectl. All you need to do is append your
summary or detail data to the current line being constructed, similar
to the way brief data is handled. Since it has to handle both headers as well as data,
there are 4 types included in the call.
$type is used as a selector to specify whether header/data and if for summary/detail
is to be constructed.
$line is a reference to the current line being constructed. Be sure to use $SEP as
a field separator so your code will correctly use collectl's --sep switch.
PrintVerbose($printHeader, $homeFlag, \$line)
Like PrintBrief, this routine is in charge of printing verbose data but is much simpler
since it doesn't have to insert code into the middle of running strings.
$printHeader is actually a flag, which when set indicate it's time to print a header.
As you can see in hello.ph, there are 2 global string worth knowing about. Both $miniFiller
and $datetime are normally set to null, but if the user specifies a time switch such as -oT,
the first will be set to the appropriate number of pad characters and the second set to the time
in the format specified by -o.
$homeFlag is a localized copy of collectl's $homeFlag which when set indicates the
display is in home mode and section separating returns should be omitted from the
output to make it more dense.
$line is a reference to the line about to be printed. Set it to the you wish to print,
rather than append to it.
UpdateHeader(\$line)
$line is a reference to the string that contains the standard header that
appears at the beginning of all collectl files. This call is made immediately before
writing out the final line of #s and so anything appended to this string will appear in
the header as its own unique line. If you do not want to add anything to the header
simply make this an empty function.
When thinking about the performance of collectl keep in mind that efficiency is one
of the main design principles and it should therefore be possible to run collectl
as a daemon on most systems with little concern for the overhead it generates.
However, given that the another design principal it to be able to collect a very broad
and deep set of data, it is worth understanding a little more about just what goes on
in collectl.
During the earlier days of collectl development a third design consideration was
minimizing the amount of storage used for raw files. At the time
disks were smaller and the amount of data collected was small enough that being more
judicious about what was saved felt like it made a difference. However, since the
inclusion of process, slab and now interrupt data along with larger disks the amount
of storage used has become less of a consideration. In other words don't spend extra
data collection cycles trying to be more selective about what is recorded if it's
going to add to the overhead.
Measuring the Overhead
If you take a closer look at the architecture you can
see that the path of minimal overhead is when collectl reads from /proc and write
it to a file. This is not an accident! In fact, on many system this has been
observed to take less than 0.1% of the CPU and this is when collecting almost
all the data collectl is capable of.
However, this also raises several important questions for consideration:
What is the overhead on a specific system?
Is there a way to reduce the overhead further?
Is collection overhead for all data types the same and if not, what is it?
In fact, these are the very questions I asked during the initial development as well as
every time I add support for new data types. In order to be able to answer these questions
quantitatively, collectl should be run with a collection interval of 0 and told to collect
8640 samples, the number of 10-second samples in a day. By timing the execution you can
then get a reasonable estimate of the daily overhead. Consider the following:
# time collectl -scdnm -i0 -c8640 -f /tmp
real 0m9.711s
user 0m7.480s
sys 0m2.140s
and you can see that collectl uses about 10 seconds out of 86400 or about 0.01% of the cpu
to collect cpu, disk, network and memory data. If we repeat the test again for just cpu
the time drops to under 4 seconds, so you can see if performance is really critical, you
can improve things by recording less data or maybe just do it less frequently. The point
is these are tradeoffs only you can make if you feel collectl is using too much resource.
So what happens if you take a different processing path and save collectl data in plot
format? This means adding the additional overhead of parsing the /proc data and performing
some basic math of the values. If we use the same command as above and include -P:
# time collectl -scdnm -i0 -c8640 -f /tmp
real 0m20.607s
user 0m17.970s
sys 0m2.580s
we can see that this takes a little over twice as much overhead even though it is
still pretty low.
One other example worth mentioning and is process monitoring overhead,
which is the highest overhead operation collectl can do and one of the reasons
it has its own monitoring interval. The overheard for collection of this type of data can
vary quite broadly depending on how may processes are running at the time and on a
system with only 138 processes look at this:
# time ./collectl.pl -sZ -i0 -c8640 -f /tmp
real 1m8.453s
user 0m54.650s
sys 0m13.430s
noting collectl is also smart enough to only look at 1/6 as many samples of process data
since that is the default relationship of process monitoring to other subsystem data.
This also leads to the mention of a way to further optimize process monitoring.
If you are monitoring a specify
set of processes, say http daemons, collectl no longer has to look at as much data
in /proc and so we now see:
# time ./collectl.pl -sZ -Zchttp -i0 -c8640 -f /tmp
real 0m5.721s
user 0m4.480s
sys 0m1.180s
In fact, if we know there are never going to be any new http process appearing
(collectl looks for new processes that match selection strings by default):
# time ./collectl.pl -sZ -Zchttp -i0 -c8640 --procopts p -f /tmp
real 0m5.080s
user 0m3.930s
sys 0m1.130s
And things get even better as you could even image monitoring these processes
at the same interval as everything else with almost no additional overhead!
Another wrinkle - process I/O statistics
Since the inclusion of process I/O stats, collectl now has to read an addition set
of data for each process, specifically /proc/pid/io, which adds over 25% to
the total process data collection load. For most users, collectl's overhead is reasonable low
enough for this extra overhead to not be a problem. Remember - we're talking about
an increase of 25% to a relatively small number. But for those concerned with this
extra overhead, a new --procopt value of I has been added to V3.6.1 which
suppresses the reading of process I/O stats.
Remember - your milage can and will vary!
The number of different combinations of switches one can measure far exceeds the scope
of this discussion - for example we haven't even talked about the pros/cons of
compression. But in conclusion, the main thing to remember is collectl's overhead is
already pretty low but if you're really concerned, measure it yourself and adjust accordingly.
By default, collectl always reports all data for all devices. However, in the
cases where there are dozens or possibly hundreds of devices such as with large
disk farms, it may be desirable to only look at those devices that are actually
doing something of interest. These are referred to as exceptions, because
their activity has crossed a level of minimal activity. The defaults for these levels
can be displayed with the -V switch or
changed to different values with the -l switch. To change one or more values
simply specify them as a string. There are currently 4 levels one can set:
SVC - Disk Service Time
IOS - Number of Disk I/O Operations
LusKBS - Lustre KBytes/Sec
LusReints - Lustre MDS Reint operations
Note that one can also specify all conditions must be met or
simply 1 must be met by adding a selection of AND (the default) or OR, respectively.
For example, to set the minimal SVC level to 50 and require both SVC and IOS
limits be reached, simply add the switch -l SVC:50.
To change both values and require only 1 be met, separate them with a
hyphen and be sure to include OR as one of the parameters such as
-l SVC:50-IOS:10-OR, noting that order is not important.
One should not confuse exceptions which are based on threshold values, with filters
which are based on the presence of explicit field values.
As of Version 3.4.2 of collectl, collectl can now detect
dynamic changes to a CPU's state. In other words, going offline or coming back online.
When one or more CPUs is indeed found to be off line, collectl will include a message in
an output header to indicate this. Furthermore, when display CPU numbers in headers, those
names will have their number changed to Xs to indicate this has occurred, such as in the
following:
[root@node02 mjs]# ./collectl.pl -scj
waiting for 1 second sample...
# *** One or more CPUs disabled ***
#<--------CPU--------><-----------------Int------------------>
#cpu sys inter ctxsw Cpu0 Cpu1 Cpu2 CpuX Cpu4 Cpu5 Cpu6 Cpu7
0 0 1051 49 1000 17 0 0 4 0 0 29
As the state changes, the headers will change accordingly. If there is a state change between
headers this won't be seen until the next headers are displayed. If displaying detail data,
one CAN tell the stat has changed. In the case of looking at only CPU data, ALL percentages for
a CPU that is offline will display as zeros. If looking at interrupts, the CPU number will be
changed to an X in the header (see Restrictions below).
When logging to a file, if any CPUs are found to be offline when collectl starts, that number
will be written to the file header in the field CPUsDis. A new flag D will also
be added to the Flags field. However, one will still see the same effects of a CPU
state change in the output during playback.
Restrictions If a CPU goes offline after collectl has started and one is logging to disk, it will not
be noted in the file header.
When monitoring process data, this header will indicate if a CPU was found to be offline at the
time collectl started as well as during processing. However, if the state changes and you're not
explicitly displaying CPU data, there will be no indication of dynamic CPU state changes reported.
If you are only monitoring interrupt data and there is a state change things will get very messy.
As users typically monitor Interrupts and CPU data at the same time it is not felt to be worth
the extra effort or processng overhead to try and accommodate this rare case.
Collectl supports 2 very basic data logging mechanisms. In the
first case it will log the data as read from /proc to a file with
the extension raw or raw.gz, depending on whether or not the
perl module Compress::Zlib.pm has been installed.
If not, one can always install compression at a later
time and collectl will happily use it the next time it is started.
One useful property of raw files is that one can play them back
using different switches/options for display or generation of
plottable files from them.
The second major form of logging is writing data to one or more tabularized,
also known as plottable
files, which have the extension tab for data associated with the core
subsystems or one of several other files for the detail data associated
with devices like cpus, disks, networks, etc.
The biggest benefit of raw files is they are very lightweight to create in that
no additional processing is performed on the data.
Since they contain the unaltered /proc data from
which collectl derives its numbers to report, it is always possible to go back
and look at the orginal data.
In some cases, there is data in the raw file that was easier to
collect than ignore and in these situations one can actually see more data than
is normally available. In fact the --grep switch is available for looking
for data in the raw files and prefacing them with timestamps, something the standard
grep command cannot do.
As their type implies, plottable files have their data in a form that is ready
to be plotted with tools like gnuplot or immediately loadable into a spreadsheet
like OpenOffice or Excel or any other tool that can read space-separated data.
When generated by collectl while it is running, this data can be read
while it is being generated making it possible to do real-time monitoring/display
of it. For situations where a tool requires data be delimited by something other
than spaces, one can change the data separator with --sep. In fact, for the
case where a tool such as rrd requires the date be in UTC format, you can even
change the timestamp format using --utc.
Logging Options
There are 2 switches you should become familiar with for logging data, noting that
you cannot write to a file and the terminal at the same time.
-f tells collectl to write a log file to the specified directory. All files
written by collectl have a predefined name format which includes the hostname and date
and in some cases the time as well. You cannot change this format but if -f does not specify
a directory name, that string is prepended to the standard output file name
-P which when used in combination with -f instructs collectl to write
a plottable file instead
As collectl continues to grow in functionality and collect more data, linux is also growing
in complexity and increasing the number of active processes as well as expanding the number
of slabs. There is a 3rd option which has been around for quite awhile but has had minimal
use or discussion and that is the -G or --group switch. When specified, this tells collectl
to write process and slab data to a second file named rawp (initially it only
contained process data). The main reason for doing so is because without this switch a typical
raw file, even when compressed, can approach 50MB or more, growing even larger as the number
of active processes grows.
While large files are nothing new to collectl, playing them back either for the purpose of drilling
into the data or to simply generate plot files can become very expensive in terms of time and CPU
load. In extreme some cases it can take tens of minutes to process a single, large raw file and even
in normal cases it will take multiple minutes. Having
collectl write to 2 separate files doesn't add any additional overhead or disk space
but can significantly reduce the playback time when you are not interested in slab or process data,
which is often the case during initial analysis.
As a data point, on my development system, single compressed collectl logs are on the order of 35MB.
When using the -G switch, it generated a pair of files where the process/slab data is about 34MB
and the file with the rest of the data is only 1MB making the raw, where all the subsystem
details are stored, very efficient to process in playback mode, taking about a minute compared to
5 minutes when that file includes slab and process data.
For most users this is all you need to know. On the other hand if
want to use collectl to feed data to other tools or perhaps log to both
raw and plot files at the same time, read on...
Logging both raw and plottable data at the same time!
The main benefit in requesting collectl to write its data in plottable form is
that data becomes available for immediate plotting without any post-processing
required, the one expense being some additional processing
overhead.
However there are a few potential limits in doing so that should be understood.
First and foremost, once a plottable file has been created the original
data from which it was created is lost forever. In many cases that is fine as many users
feel there is really no need to go back to the original source.
However, one often collects summary data because
that is what they are interested in, but then later decides they want to look at the details.
This can be easily done by just replaying the raw file and requesting details be
displayed or (re)written to a plottable file. If the raw file had not
been generated, this option would not be possible.
A second limitation with plottable
data files is that one cannot easily examine the data by timeframes
and when there are multiple data files involved,
it is not easy to look at all the data together as time-oriented samples without plotting it.
It is always possible to write a script that merges this data together, but that
functionality is natively built into collectl when used in playback mode.
Finally, there are times when one might wish to go back and look at non-normalized
data, for example if one has 3 processes created over a 10 second period
collectl will report a rate of 0 process creations/second because it would round down
and the only way to
see what really happened is to play the data back with -on, which tells collectl
not to normalize the data and will therefore tell you the value of
the counter not its rate.
In most cases none of these restrictions should be a concern, but there
may be occasions in which they are and that is where the --rawtoo switch
comes in. When specified in conjunction with -P,
collectl will generate raw data in addition to the plottable
data, making it possible to go back to the source if/when necessary.
The only real overhead is the amount of disk space required since the raw data
is already sitting in a buffer and ready to be written.
If the plottable files are being generated in uncompressed format,
the size of the compressed raw file becomes even less significant.
Exported output, the 3rd type of file
We finally come to a third type of output, intended primarily for feeding
collectl data to other programs, and that is exported output. There are
currently a variety of types of exports delivered with collectl though
only three are capable of generating local data files and their use complicates the
picture. To better understand how logging works in the context of --export,
see the description of how they work and in particular
how collectl decides where and when to do logging.
The first is the s-expression.
S-expressions have been around for many years having their earliest roots in
programming languages such as Lisp and Scheme, as described in the Wikipedia
and offer a semi-structured mechanism for the representation of data. One such
environment in which they are heavily used is
supermon
and by providing a mechanism for collectl to
write s-expressions, one can more easily supply data to supermon or any other
tools that might wish to consume it in close to real-time.
The actual contents of the s-expressions will be driven by the subsystems for which
data is being collected.
The second type is the L-expression which is something that has been
invented purely for collectl as an alternative form of output which in some
environments can be easier to parse than the more complex s-sepression.
The third and newest type, named gexpr, writes its output over a UDP
socket in a format understood by the ganglia
monitoring daemon, gmond. You can read more about this here.
If you are a little confused, and you probably should be, try experimenting with
various combinations of switches and see which files get generated.
The overhead
So what is the overhead associated with all this logging? From the perspective of
CPU load it can be quite minimal since in most cases the data is already in hand and
all that needs to be done is to write it out to one or more additional files, something that
is a fairly low-overhead operation on Linux systems. If this is really a concern,
measure it yourself. It you want to see how much disk space involved
just examine the sizes of the file(s) created during the performance tests and see for yourself.
It is now possible to monitor nvidia GPUs by including the option --import nvidia
with the collectl command. Naturally if you've recorded nvidia data you'll need to
include this switch on playback as well.
The nvidia module currently reports 4 variables: temperature, fan speed and GPU/memory
utilization. Other than temperature which is in degrees C, the other 3 are reported as
percentages.
This module supports all 3 collectl formatting modes, specifically brief,
verbose and detail. Since there is nothing more of interest to report in
verbose mode, both it and brief mode report the same thing. As with all other collectl
output modes, brief/verbose report averages (since totals would make no sense) and in
detail mode one can see values for each GPU. However if there is only 1 GPU present, all
three modes report the same results.
There is one key thing to be aware of with respect to what data is reported. Since it
normally would not be possible to report data collected at different intervals in a
consistent line of brief format, and by default this module only collects data once a minute,
a technique introduced in the import misc.ph is used and that is to always report
the more recent value collected even if it changes at a different frequency! In other words,
just because a value was reported with a specific timestamp it does not necessarily mean
it was collected at that exact time.
The following example illustrates several of things worth noting:
As described in the section on performance, it actually takes over a second
to read the data from a GPU and that is visually illustrated above because one can see
gaps in the data being reported. Since we're running collectl interactively, it wants
to report once a second but when it reads the GPU data it misses an interval.
More specifically, the data is actually read during the interval that
started 06:23:41 and since it took over a second the next interval didn't start until 2
seconds later! Then at 06:23:43 and 06:23:44, the data read earlier is reported. At
06:23:45 data is again read from the GPU and since we miss an interval, the next one
occurs are 06:23:47.
Performance
The interface actually calls the nvidia command nvidia-smi to query the GPUs, which
unfortunately is not very lightweight, taking about 1.3 seconds to query the GPU.
Recalling that there are 86400 seconds in a day and 84 seconds is a 0.1% load, one would
not want to run this commandevery 10 second monitoring cycle! By default, a monitoring
interval of 1 minute has been chosen which will still use a little over 2% of a CPU, but
it is also felt that most of the time one chooses this option to see what is happening
within the GPU and one may miss important changes with less frequent monitoring.
However, as with other --import modules, one can see in the example above that it
is easy to change the monitoring frequency. In other words to monitor the GPU at 5 minute
intervals, use --import nvidia,i=300.
Version 3.2.1 of collectl introduces support for /proc/buddyinfo, which shows the distribution
of memory fragments where the size of each fragment is a power of 2 pages. The memory is categorized
by node, depending on the system architecture and then subcategorized by the type of memory, referred
as a zone. For example, the /proc/buddyinfo might look like:
Running collectl with -sb --verbose would produce a single line of output that shows the totals
of each column. For example, taking 1 second samples with timestamps included:
Where things get interesting is in brief mode. The challege here is to show the maximum amount of
information in the least amount of space, thus allowing you to look at other information on the same line
as well. To better understand the methodology chosen for this, think in terms of base 36.
However, instead of mapping each character to a number from 0 to 35, we're going to do something very
different. Since we're mainly interested in seeing what's happening as the numbers of fragments shrink:
map the 0 through 9 to the characters 0 through 9
map the numbers 10-99 to the characters a through i by dividing these values by 10
fragments in the range of 100-999 will be divided by 100 and mapped to the letters j through r
this only leaves us with the 8 letters s through z and being the coarsest
will be mapped to powers of 10 by essentially dividing the value by 1000 and taking the log10() of it
As a final example of what this looks like when combined with other data, remembering you can choose
virtually any combination of subsystem for your display both in collection and playback modes:
collectl-3.6.9/ARTISTIC 0000644 0001750 0001750 00000013733 12230241726 012651 0 ustar mjs mjs The "Artistic License"
Preamble
The intent of this document is to state the conditions under which a
Package may be copied, such that the Copyright Holder maintains some
semblance of artistic control over the development of the package,
while giving the users of the package the right to use and distribute
the Package in a more-or-less customary fashion, plus the right to make
reasonable modifications.
Definitions:
"Package" refers to the collection of files distributed by the
Copyright Holder, and derivatives of that collection of files
created through textual modification.
"Standard Version" refers to such a Package if it has not been
modified, or has been modified in accordance with the wishes
of the Copyright Holder as specified below.
"Copyright Holder" is whoever is named in the copyright or
copyrights for the package.
"You" is you, if you're thinking about copying or distributing
this Package.
"Reasonable copying fee" is whatever you can justify on the
basis of media cost, duplication charges, time of people involved,
and so on. (You will not be required to justify it to the
Copyright Holder, but only to the computing community at large
as a market that must bear the fee.)
"Freely Available" means that no fee is charged for the item
itself, though there may be fees involved in handling the item.
It also means that recipients of the item may redistribute it
under the same conditions they received it.
1. You may make and give away verbatim copies of the source form of the
Standard Version of this Package without restriction, provided that you
duplicate all of the original copyright notices and associated disclaimers.
2. You may apply bug fixes, portability fixes and other modifications
derived from the Public Domain or from the Copyright Holder. A Package
modified in such a way shall still be considered the Standard Version.
3. You may otherwise modify your copy of this Package in any way, provided
that you insert a prominent notice in each changed file stating how and
when you changed that file, and provided that you do at least ONE of the
following:
a) place your modifications in the Public Domain or otherwise make them
Freely Available, such as by posting said modifications to Usenet or
an equivalent medium, or placing the modifications on a major archive
site such as uunet.uu.net, or by allowing the Copyright Holder to include
your modifications in the Standard Version of the Package.
b) use the modified Package only within your corporation or organization.
c) rename any non-standard executables so the names do not conflict
with standard executables, which must also be provided, and provide
a separate manual page for each non-standard executable that clearly
documents how it differs from the Standard Version.
d) make other distribution arrangements with the Copyright Holder.
4. You may distribute the programs of this Package in object code or
executable form, provided that you do at least ONE of the following:
a) distribute a Standard Version of the executables and library files,
together with instructions (in the manual page or equivalent) on where
to get the Standard Version.
b) accompany the distribution with the machine-readable source of
the Package with your modifications.
c) give non-standard executables non-standard names, and clearly
document the differences in manual pages (or equivalent), together
with instructions on where to get the Standard Version.
d) make other distribution arrangements with the Copyright Holder.
5. You may charge a reasonable copying fee for any distribution of this
Package. You may charge any fee you choose for support of this
Package. You may not charge a fee for this Package itself. However,
you may distribute this Package in aggregate with other (possibly
commercial) programs as part of a larger (possibly commercial) software
distribution provided that you do not advertise this Package as a
product of your own. You may embed this Package's interpreter within
an executable of yours (by linking); this shall be construed as a mere
form of aggregation, provided that the complete Standard Version of the
interpreter is so embedded.
6. The scripts and library files supplied as input to or produced as
output from the programs of this Package do not automatically fall
under the copyright of this Package, but belong to whoever generated
them, and may be sold commercially, and may be aggregated with this
Package. If such scripts or library files are aggregated with this
Package via the so-called "undump" or "unexec" methods of producing a
binary executable image, then distribution of such an image shall
neither be construed as a distribution of this Package nor shall it
fall under the restrictions of Paragraphs 3 and 4, provided that you do
not represent such an executable image as a Standard Version of this
Package.
7. C subroutines (or comparably compiled subroutines in other
languages) supplied by you and linked into this Package in order to
emulate subroutines and variables of the language defined by this
Package shall not be considered part of this Package, but are the
equivalent of input as in Paragraph 6, provided these subroutines do
not change the language in any way that would cause it to fail the
regression tests for the language.
8. Aggregation of this Package with a commercial distribution is always
permitted provided that the use of this Package is embedded; that is,
when no overt attempt is made to make this Package's interfaces visible
to the end user of the commercial distribution. Such use shall not be
construed as a distribution of this Package.
9. The name of the Copyright Holder may not be used to endorse or promote
products derived from this software without specific prior written permission.
10. THIS PACKAGE IS PROVIDED "AS IS" AND WITHOUT ANY EXPRESS OR
IMPLIED WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED
WARRANTIES OF MERCHANTIBILITY AND FITNESS FOR A PARTICULAR PURPOSE.
The End
collectl-3.6.9/envrules.std 0000644 0001750 0001750 00000005034 12230241726 014056 0 ustar mjs mjs # This block at the beginning don't have rules but are there to make parsing
# this file every so slightly faster, noting the commas are really only
# needed to make visual parsing easier.
>ProLiant BL460c G1,ProLiant BL460c G6,ProLiant BL490c G6<
>ProLiant BL685c G1,ProLiant BL2x220c G5<
>ProLiant DL145 G2<
>ProLiant DL360 G3,ProLiant DL380 G3,ProLiant DL380 G5<
>ProLiant DL385 G5,ProLiant DL585 G5,ProLiant DL585 G7<
>ProLiant ML370 G6<
# These are the real rules and the headers much EXACTLY match everything to the
# right of the ": " returned by the command "dmidecode|grep -m1 'Product Name'"
# for example: "Product Name: ProLiant BL460c G1"
>ProLiant DL185 G5<
[pre]
/CPU(\d) Diode/CPUTemp$1/
/Power Ambient/ATemp/
/Front Panel Temp/FPTemp/
>ProLiant DL160 G5<
[pre]
/ ROTOR/R/
/Rear Ambient/RTemp/
/PCI Ambient/ATemp/
/CPU(\d) Dmn (\d) Temp/Temp$1-$2/
>ProLiant DL785<
[pre]
/Zone /Z/
>SE4208<
[pre]
/^CPU/CTemp/
/^System/STemp/
/^FAN6CPU/Fan6C/
# There is a lot of remappings to get SL170h temperatures into fixed format as follows
# CPU1 Temp->CTemp1 CPU2 Temp->CTemp2 CPU1 DIMM->CTemp1D CPU2 DIMM->CTemp2D
# CPU2_DIMM6G->CTemp6G CPU2_DIMM5B->CTemp5B CPU2_DIMM4E->CTemp4E
# CPU2_DIMM3H->CTemp3H CPU2_DIMM2C->CTemp2C CPU2_DIMM1F->CTemp1F
# Node Exit Amb->ETemp Front I/O Amb->FTemp HDD BP Amb->HTemp
# Node Inlet Amb->ITemp PCIE local Temp->LTemp Northbridge Amb->NTemp PSU BP->PTemp
>ProLiant SL170h G6,ProLiant SL2x170z G6<
[pre]
/^CPU(\d) DIMM Amb/CTemp$1D/
/^Node //
/Amb/Temp/
/DIMM(.*)/Temp$1/
/.*Internal/Temp/
/^PCIE l/L/
# This on also has a lot of remapping
>ProLiant DL160 G6,ProLiant DL160se G6<
[pre]
/_INLET/I/
/_OUTLET/O/
/Rear Board/xTemp/
/^Front Board/BTemp/
/^Front dimm/FTemp/
/^Rear dimm/RTemp/
/^PCI outlet/PTemp/
/^IOH outlet/ITemp1/
/^Front IOH/ITemp2/
/^Inlet ambient/ATemp/
/ sensor/ Temp/
[post]
/^BTemp/BTempF/
/^xTemp/BTempR/
[ignore]
/^Fan Redundant/
>ProLiant SL230s Gen8,ProLiant SL250s Gen8<
[pre]
/^\d*-//
/Inlet Ambient/ATemp/
/CPU/CTemp/
/Chipset/CTemp/
/.*P(\d) DIMM (\d).*/$1Temp$2D/
/HD Max/mTemp/
/HD C.*/cTemp/
/HD z.*/zTemp/
/VR P(\d)Mem Zone/VTemp$1MZ/
/VR P(\d) ([MZ]).*/VTemp$1$2/
/VR P(\d)/VTemp$1/
/SuperCap.*/STemp/
/LOM Card/lTemp/
/LOM/LTemp/
/PCI 2 Zone/pTemp/
/PCI 2/PTemp/
/PS Board/uTemp/
/Sys Ex.*/STemp/
/Coprocessor (\d)/GTemp$1/
[post]
/^STemp/STempE/
/^cTemp/HTempC/
/^mTemp/HTempM/
/^zTemp/HTempZ/
/^lTemp/LTempC/
/^pTemp/PTempZ/
/^uTemp/PTempS/
>SE1170s<
[pre]
/Fan(\d)a/F$1a/
/CPU(\d)_DIMM(.*)/$1Temp$2/
/HDD BP(\d).*/HTemp$1/
/PCIE.*/PTemp/
/Front.*/ATemp/
collectl-3.6.9/COPYING 0000644 0001750 0001750 00000000231 12230241726 012524 0 ustar mjs mjs collectl may be copied only under the terms of either the Artistic License
or the GNU General Public License, which are also located in this directory.
collectl-3.6.9/collectl.pl 0000755 0001750 0001750 00000746331 12230241726 013654 0 ustar mjs mjs #!/usr/bin/perl -w
# Copyright 2003-2013Hewlett-Packard Development Company, L.P.
#
# collectl may be copied only under the terms of either the Artistic License
# or the GNU General Public License, which may be found in the source kit
# debug
# 1 - print interesting stuff
# 2 - print Interconnect specific checks (mostly Infiniband)
# 4 - show each line processed by record(), replaces -H
# 8 - print lustre specific checks
# 16 - print headers of each file processed
# 32 - skip call to dataAnalyze during interactive processing
# 64 - socket processing
# 128 - show collectl.conf processing
# 256 - show detailed pid processing (this generates a LOT of output)
# 512 - show more pid details, specifically hash contents
# NOTE - output from 256/512 are prefaced with %%% if from collectl.pl
# and ### if from formatit.ph
# 1024 - show list of SLABS to be monitored
# 2048 - playback preprocessing
# 4096 - report pidNew() management of pidSkip{}
# 8192 - show creation of RAW, PLOT and files
# debug tricks
# - use '-d36' to see each line of raw data as it would be logged but not
# generate any other output
# Equivalent Utilities
# -s c mpstat, iostat -c, vmstat
# -s C mpstat
# -s d/D iostat -d
# -s f/F nfsstat -c/s [c if -o C]
# -s i sar -v
# -s m sar -rB, free, vmstat (note - sar does pages by pagesizsie NOT bytes)
# -s n/N netstat -i
# -s s sar -n SOCK
# -s y/Y slabtop
# -s Z ps or top
# Subsystems
# b - buddy
# c - cpu
# d - disks
# E - environmental
# i - inodes (and other file stuff)
# f - NFS
# l - lustre
# m - memory
# n - network
# s - socket
# t - tcp
# x - interconnect
# Z - processes (-sP now available but -P taken!)
use POSIX;
use Config;
use English;
use 5.008000;
use Getopt::Long;
Getopt::Long::Configure ("bundling");
Getopt::Long::Configure ("no_ignore_case");
Getopt::Long::Configure ("pass_through");
use File::Basename;
use Time::Local;
use IO::Socket;
use IO::Select;
$Cat= '/bin/cat';
$Grep= '/bin/grep';
$Egrep= '/bin/egrep';
$Ps= '/bin/ps';
$Rpm= '/bin/rpm';
$Lspci= '/sbin/lspci';
$Lctl= '/usr/sbin/lctl';
$Dmidecode= '/usr/sbin/dmidecode';
$ReqDir= '/usr/share/collectl'; # may not exist
%TopProcTypes=qw(vsz '' rss '' syst '' usrt '' time '' accum '' rkb '' wkb '' iokb ''
rkbc '' wkbc '' iokbc '' ioall '' rsys '' wsys '' iosys ''
iocncl '' majf '' minf '' flt '' pid '' cpu '' thread '' vctx '' nctx '');
%TopSlabTypes=qw(numobj '' name '' actobj '' objsize '' numslab '' objslab '' totsize '' totchg '' totpct '');
# Constants and removing -w warnings
$miniDateFlag=0;
$PageSize=0;
$Memory=$Swap=$Hyper=$Distro=$ProductName='';
$CpuVendor=$CpuMHz=$CpuCores=$CpuSiblings=$CpuNodes='';
$PidFile='/var/run/collectl.pid'; # default, unless --pname
$PQuery=$PCounter=$VStat=$VoltaireStats=$IBVersion=$HCALids=$OfedInfo='';
$numBrwBuckets=$cfsVersion=$sfsVersion='';
$Resize=$IpmiCache=$IpmiTypes=$ipmiExec='';
$i1DataFlag=$i2DataFlag=$i3DataFlag=0;
$lastSecs=$interval2Print=0;
$diskRemapFlag=$diskChangeFlag=$cpuDisabledFlag=$cpusDisabled=$cpusEnabled=$noCpusFlag=0;
$boottime=0;
# only used once here, but set in formatit.ph
our %netSpeeds;
# Find out ASAP if we're linux or WNT based as well as whether or not XC based
$PcFlag=($Config{"osname"}=~/MSWin32/) ? 1 : 0;
$XCFlag=(!$PcFlag && -e '/etc/hptc-release') ? 1 : 0;
# If we ever want to write something to /var/log/messages, we need this which
# we obviously can't include on a pc.
require "Sys/Syslog.pm" if !$PcFlag;
# Always nice to know if we're root
$rootFlag=(!$PcFlag && `whoami`=~/root/) ? 1 : 0;
$SrcArch= $Config{"archname"};
$Version= '3.6.9-1';
$Copyright='Copyright 2003-2013 Hewlett-Packard Development Company, L.P.';
$License= "collectl may be copied only under the terms of either the Artistic License\n";
$License.= "or the GNU General Public License, which may be found in the source kit";
# get the path to the exe from the program location, noting different handling
# of path resolution for XC and non-XC, noting if a link and not XC, we
# need to follow it, possibly multiple times! Furthermore, if the link is
# a relative one, we need to prepend with the original program location or
# $BinDir will be wrong.
if (!$XCFlag)
{
$link=$0;
$ExeName='';
until($link eq $ExeName)
{
$ExeName=$link; # possible exename
$link=(!defined(readlink($link))) ? $link : readlink($link);
}
}
else
{
$ExeName=(!defined(readlink($0))) ? $0 : readlink($0);
$ExeName=dirname($0).'/'.$ExeName if $ExeName=~/^\.\.\//;
}
$BinDir=dirname($ExeName);
$Program=basename($ExeName);
$Program=~s/\.pl$//; # remove extension for production
# Note that if someone redirects stdin or runs it out of a script it will look like
# we're in the background. We also need to know if STDOUT connected to a terminal.
if (!$PcFlag)
{
$MyDir=`pwd`;
$Cat= 'cat';
$Sep= '/';
$backFlag=(getpgrp()!=tcgetpgrp(0)) ? 1 : 0;
$termFlag= (-t STDOUT) ? 1 : 0;
}
else
{
$MyDir=`cd`;
$Cat= 'type';
$Sep= '\\';
$backFlag=0;
$termFlag=0;
}
chomp $MyDir;
# This is a little messy. In playback mode of process data, we want to use
# usernames instead of UIDs, so we need to know if we need to know if it's
# the same node and hence we need our name. This could be different than $Host
# which was recorded with the data file and WILL override in playback mode.
# We also need our host name before calling initRecord() so we can log it at
# startup as well as for naming the logfile.
$myHost=($PcFlag) ? `hostname` : `/bin/hostname`;
$myHost=(split(/\./, $myHost))[0];
chomp $myHost;
$Host=$myHost;
# may be overkill, but we want to throttle max errors/day to prevent runaway.
$zlibErrors=0;
# These variables only used once in this module and hence generate warnings
undef @dskOrder;
undef @netOrder;
undef @lustreCltDirs;
undef @lustreCltOstDirs;
undef @lustreOstSubdirs;
undef %playbackSettings;
$recHdr1=$recHeader=$miniDateTime=$miniFiller=$DaemonOptions='';
$OstNames=$MdsNames=$LusDiskNames=$LusDiskDir='';
$NumLustreCltOsts=$NumLusDisks=$MdsFlag=0;
$NumSlabs=$SlabGetProc=$newSlabFlag=0;
$wideFlag=$coreFlag=$newRawSlabFlag=0;
$totalCounter=$separatorCounter=0;
$NumCpus=$HZ='';
$NumOst=$NumBud=0;
$FS=$ScsiInfo=$HCAPortStates='';
$SlabVersion=$XType=$XVersion='';
$dentryFlag=$inodeFlag=$filenrFlag=$allThreadFlag=$procCmdWidth=0;
$clr=$clscr=$cleol=$home='';
$dskIndexNext=$netIndexNext=0;
# This tells us we have not yet made our first pass through the data
# collection loop and gets reset to 0 at the bottom.
$firstPass=1;
# Check the switches to make sure none requiring -- were specified with -
# since getopts doesn't! Also save the list of switches we were called with.
$cmdSwitches=preprocSwitches();
# These are the defaults for interactive and daemon subsystems
$SubsysDefInt='cdn';
$SubsysDefDaemon='bcdfijlmnstx';
# We want to load any default settings so that user can selectively
# override them. We're giving these starting values in case not
# enabled in .conf file. We later override subsys if interactive
$SubsysDef=$SubsysCore=$SubsysDefDaemon;
$Interval= 10;
$Interval2= 60;
$Interval3= 120;
$LimSVC= 30;
$LimIOS= 10 ;
$LimLusKBS= 100;
$LimLusReints=1000;
$LimBool= 0;
$Port= 2655;
$Timeout= 10;
$MaxZlibErrors=20;
$LustreSvcLunMax=10;
$LustreMaxBlkSize=512;
$LustreConfigInt=1;
$InterConnectInt=900;
$TermHeight=24;
$DefNetSpeed=10000;
$IbDupCheckFlag=1;
$TimeHiResCheck=1;
$PasswdFile='/etc/passwd';
$umask='';
$DiskMaxValue=-1; # disabled
# NOTE - the following line should match what is in collectl.conf. If uncommented there, it will be replaced
$DiskFilter='cciss/c\d+d\d+ |hd[ab] | sd[a-z]+ |dm-\d+ |xvd[a-z] |fio[a-z]+ | vd[a-z]+ |emcpower[a-z]+ |psv\d+ ';
$DiskFilterFlag=0; # only set when filter set in collectl.conf
$ProcReadTest='yes';
# Standard locations
$SysIB='/sys/class/infiniband';
# These aren't user settable but are needed to build the list of ALL valid
# subsystems
$SubsysDet= "BCDEFJLMNTXYZ";
$SubsysExcore='y';
# These are the subsystems allowed in brief mode
$BriefSubsys="bcdfijlmnstx";
# And the default environmentals
$envOpts='fpt';
$envRules='';
$envDebug=0;
$envTestFile='';
$envFilt=$envRemap='';
$hiResFlag=0; # must be initialized before ANY calls to error/logmsg
$configFile='';
$ConfigFile='collectl.conf';
$daemonFlag=$debug=$formatitLoaded=0;
GetOptions('C=s' => \$configFile,
'D!' => \$daemonFlag,
'd=i' => \$debug,
'config=s' => \$configFile,
'daemon!' => \$daemonFlag,
'debug=i' => \$debug
) or error("type -h for help");
# if config file specified and a directory, prepend to default name otherwise
# use the whole thing as the name.
$filename='';
$configFile.="/$ConfigFile" if $configFile ne '' && -d $configFile;;
loadConfig();
# Very unlikely but I hate programs that silently exit. We have to figure out
# where formatit.ph lives, out first choice always being '$BinDir'.
$BinDir=dirname($ExeName);
if (!-e "$BinDir/formatit.ph" && !-e "$ReqDir/formatit.ph")
{
# Let's not get too carried away for something that probably won't ever happen on a PC,
# but there's no point displaying $ReqDir since it in unix format and will never exist!
my $msg=sprintf("can't find 'formatit.ph' in '$BinDir'%s. Corrupted installation!", !$PcFlag ? " OR '$ReqDir'" : '');
print "$msg\n"; # can't call logmsg() before formatit.ph not yet loaded
logsys($msg,1); # force it because $filename not yet set
exit(1);
}
# Now that we've loaded collectl.conf and have possibly reset '$ReqDir', it's time to
# load it, changing $ReqDir to $BinDir if we find it there.
$ReqDir=$BinDir if -e "$BinDir/formatit.ph";
print "BinDir: $BinDir ReqDir: $ReqDir\n" if $debug & 1;
# Load include files and optional PMs if there
require "$ReqDir/formatit.ph";
$zlibFlag= (eval {require "Compress/Zlib.pm" or die}) ? 1 : 0;
$hiResFlag= (eval {require "Time/HiRes.pm" or die}) ? 1 : 0;
$diskRemapFlag=(eval {require "$ReqDir/diskremap.ph" or die}) ? 1 : 0;
$formatitLoaded=1;
# These can get overridden after loadConfig(). Others can as well but this is
# a good place to reset those that don't need any further manipulation
$limSVC=$LimSVC;
$limIOS=$LimIOS;
$limBool=$LimBool;
$limLusKBS=$LimLusKBS;
$limLusReints=$LimLusReints;
$termHeight=$TermHeight;
# On LINUX and only if associated with a terminal in the foreground and we can find 'resize',
# use the value of LINES to set the terminal height
if (!$PcFlag && !$daemonFlag && !$backFlag && $Resize ne '' && $termFlag && defined($ENV{TERM}) && $ENV{TERM}=~/xterm/)
{
# IF the user typed a CR after collectl started but before it started, flush input buffer
my $selTemp=new IO::Select(STDIN);
while ($selTemp->can_read(0))
{ my $temp=; }
$selTemp->remove();
`$Resize`=~/LINES.*?(\d+)/m;
$termHeight=$1;
}
# let's also see if there is a terminal attached. this is currently only
# an issue for 'brief mode', but we may need to know some day for other
# reasons too. but PCs can only run on a terminal...
$termFlag=(open TMP, " \$alignFlag,
'A=s' => \$address,
'address=s' => \$address,
'c=i' => \$count,
'count=i' => \$count,
'f=s' => \$filename,
'filename=s' => \$filename,
'F=i' => \$flush,
'flush=i' => \$flush,
'G!' => \$groupFlag,
'group!' => \$groupFlag,
'tworaw!' => \$tworawFlag,
'home!' => \$homeFlag,
'i=s' => \$userInterval,
'interval=s' => \$userInterval,
'h!' => \$hSwitch,
'help!' => \$hSwitch,
'iosize!' => \$ioSizeFlag,
'l=s' => \$limits,
'limits=s' => \$limits,
'L=s' => \$lustreSvcs,
'lustsvcs=s' => \$lustreSvcs,
'm!' => \$msgFlag,
'messages!' => \$msgFlag,
'o=s' => \$userOptions,
'options=s' => \$userOptions,
'N!' => \$niceFlag,
'nice!' => \$niceFlag,
'nohup!' => \$nohupFlag,
'passwd=s' => \$passwdFile,
'p=s' => \$playback,
'playback=s' => \$playback,
'P!' => \$plotFlag,
'quiet!' => \$quietFlag,
'plot!' => \$plotFlag,
'r=s' => \$rollLog,
'rolllogs=s' => \$rollLog,
'R=s' => \$runTime,
'runtime=s' => \$runTime,
's=s' => \$userSubsys,
'sep=s' => \$SEP,
'stats!' => \$statsFlag,
'statopts=s' => \$statOpts,
'subsys=s' => \$userSubsys,
'top=s' => \$topOpts,
'utc!' => \$utcFlag,
'umask=s' => \$umask,
'utime=i' => \$utimeMask,
'v!' => \$vSwitch,
'version!' => \$vSwitch,
'V!' => \$VSwitch,
'showdefs!' => \$VSwitch,
'w!' => \$wideFlag,
'x!' => \$xSwitch,
'helpextend!'=> \$xSwitch,
'X!' => \$XSwitch,
'helpall!' => \$XSwitch,
'slabfilt=s' => \$slabFilt,
'procfilt=s' => \$procFilt,
'all!' => \$allFlag,
'comment=s' => \$comment,
'cpuopts=s' => \$cpuOpts,
'dskfilt=s' => \$dskFilt,
'dskopts=s' => \$dskOpts,
'export=s' => \$export,
'from=s' => \$from,
'thru=s' => \$thru,
'headerrepeat=i'=> \$headerRepeat,
'hr=i' => \$headerRepeat,
'import=s' => \$import,
'intfilt=s' => \$intFilt,
'lustopts=s' => \$lustOpts,
'memopts=s' => \$memOpts,
'netfilt=s' => \$netFilt,
'netopts=s' => \$netOpts,
'nfsopts=s' => \$nfsOpts,
'nfsfilt=s' => \$nfsFilt,
'envopts=s' => \$userEnvOpts,
'envrules=s' => \$envRules,
'envdebug!' => \$envDebug,
'envtest=s' => \$envTestFile,
'envfilt=s' => \$envFilt,
'envremap=s' => \$envRemap,
'extract=s' => \$extract,
'grep=s' => \$grepPattern,
'offsettime=s' => \$offsetTime,
'pname=s' => \$pname,
'procanalyze!' => \$procAnalFlag,
'procopts=s' => \$procOpts,
'procstate=s' => \$procState,
'rawtoo!' => \$rawtooFlag,
'rawdskfilter=s'=> \$rawDskFilter,
'rawdskignore=s'=> \$rawDskIgnore,
'rawnetfilter=s'=> \$rawNetFilter,
'rawnetignore=s'=> \$rawNetIgnore,
'runas=s' => \$runas,
'showsubsys!' => \$showSubsysFlag,
'showoptions!' => \$showOptionsFlag,
'showsubopts!' => \$showSuboptsFlag,
'showtopopts!' => \$showTopoptsFlag,
'showheader!' => \$showHeaderFlag,
'showcolheaders!' =>\$showColFlag,
'showslabaliases!' =>\$showSlabAliasesFlag,
'showrootslabs!' =>\$showRootSlabsFlag,
'slabanalyze!' => \$slabAnalFlag,
'slabopts=s' => \$slabOpts,
'tcpfilt=s' => \$tcpFilt,
'verbose!' => \$verboseFlag,
'vmstat!' => \$vmstatFlag,
'whatsnew!' => \$whatsnewFlag,
'xopts=s' => \$xOpts,
) or error("type -h for help");
# This needs to be done BEFORE processing --pname since we end up changing $PidFile
if ($runas ne '')
{
error("canot use --runas without -D") if !$daemonFlag;
# temporariluy disable daemon mode in debug mode so we can see messages on terminal.
$daemonFlag=0 if $debug;
my ($runasUser,$runasGroup)=split(/:/, $runas);
error("--runas must at least specify a user") if $runasUser eq '';
if ($runasUser!~/^\d+$/)
{
$runasUid=(split(/:/, `grep $runasUser /etc/passwd`))[2];
error("can't find '$runasUser' in /etc/passwd. Consider UID.") if !defined($runasUid);
}
if (defined($runasGroup) && $runasGroup!~/^\d+$/)
{
$runasGid=(split(/:/, `grep $runasGroup /etc/group`))[2];
error("can't find '$runasGroup' in /etc/group. Consider GID.") if !defined($runasGid);
}
$runasUid=$runasUser if $runasUser=~/^\d+/;
$runasGid=$runasGroup if defined($runasGroup) && $runasGroup=~/^\d+/;
# let's make sure the owner/group of the logging directory match
my $logdir=dirname("$filename/collectl");
($uid,$gid)=(stat($logdir))[4,5];
error("Ownership of '$logdir' doesn't match '$runas'")
if ($uid!=$runasUid) || (defined($runasGid) && $gid!=$runasGid);
# Daemon also means --nohup
$daemonFlag=$nohupFlag=1;
}
if ($pname ne '')
{
# We need to include switches because collectl-generic expects to find them in the process name
$0="collectl-$pname $cmdSwitches";
$PidFile=~s/collectl\.pid/collectl-$pname.pid/;
print "Set PName to collectl-$pname\n" if $debug & 1;
}
# O p e n A S o c k e t ?
# It's real important we do this as soon as possible because if someone runs
# us in 'client' mode, and an error occurs the server would still be hanging
# around waiting for someone to connect to that socket! This way we connect,
# report the error and exit and the caller is able to detect it.
$sockFlag=$clientFlag=$serverFlag=0;
if ($address ne '')
{
if ($address=~/\./)
{
($address,$port,$timeout)=split(/:/, $address);
$port=$Port if !defined($port) || $port eq '';
$Timeout=$timeout if defined($timeout);
$socket=new IO::Socket::INET(
PeerAddr => $address,
PeerPort => $port,
Proto => 'tcp',
Timeout => $Timeout) or
error("Could not create socket to $address:$port. Reason: $!")
if !defined($socket);
print "Socket opened on $address:$port\n" if $debug & 64;
push @sockets, $socket;
$clientFlag=1;
}
elsif ($address=~/^server/i)
{
($port, $port, $options)=split(/:/, $address, 3);
$port=$Port if !defined($port);
# Note this socket uses a different variable because when we get
# a connection we use the SAME one to talk to client as we do in
# client mode.
$sockServer = new IO::Socket::INET(
Type=>SOCK_STREAM,
Reuse=>1, Listen => 1,
LocalPort => $port) ||
error("Could not create local socket on port $port Reason: $!");
print "Server socket opened on port $port\n" if $debug & 64;
$select=new IO::Select($sockServer);
$serverFlag=1;
}
else
{
logmsg('F', 'Invalid -A option');
}
$sockFlag=1;
}
# I'm probably the only one who cares, but in -p --top -s, don't default
# to a --hr of 5, use 20
$headerRepeat=20 if $topFlag && $playback ne '' && $headerRepeat==5;
# If we used to trap these before we opened the socket, but then we couldn't
# send the message back to the called cleanly!
if ($sockFlag)
{
error("-p not allowed with -A") if $playback ne '';
error("-D not allowed with -A address") if $daemonFlag && !$serverFlag;
}
# Since the output could be intended for a socket (called from colgui/colmux),
# we need to do after we open the socket.
error() if $hSwitch;
showVersion() if $vSwitch;
showDefaults() if $VSwitch;
extendHelp() if $xSwitch;
showSubsys() if $showSubsysFlag;
showOptions() if $showOptionsFlag;
showSubopts() if $showSuboptsFlag;
showTopopts() if $showTopoptsFlag;
showSlabAliases($slabFilt) if $showSlabAliasesFlag || $showRootSlabsFlag;
whatsnew() if $whatsnewFlag;
if ($XSwitch)
{
extendHelp(1);
showSubsys(1);
showOptions(1);
showSubopts(1);
showTopopts(1);
printText("$Copyright\n");
printText("$License\n");
exit(0);
}
# in playback mode all we're really doing is verifying the options
setNFSFlags($nfsFilt);
if ($vmstatFlag)
{
error("can't mix --vmstat with --export") if $vmstatFlag && $export ne '';
error("can't mix --vmstat with --all") if $vmstatFlag && $allFlag;
$export='vmstat';
}
error("can't use --export with --verbose") if $verboseFlag && $export ne '';
error("can't use -P with --verbose") if $verboseFlag && $plotFlag;
error("can't use -f with --verbose") if $verboseFlag && $filename ne '';
error("--utime requires HiRes timer") if $utimeMask && !$hiResFlag;
error("--utime requires -f") if $utimeMask && $filename eq '';
error("max value for --utime is 7") if $utimeMask>7;
# --all is shortcut for all summary data
if ($allFlag)
{
error("can't mix -s with -all") if $userSubsys ne '';
$userSubsys="$SubsysCore$SubsysExcore";
$userSubsys=~s/y//;
}
# As part of the conversion to getopt::long, we need to know the actual switch
# values as entered by the user. Those are stored in '$userXXX' and then that
# is treated as one used to handle opt_XXX.
$options= $userOptions;
$interval=($userInterval ne '') ? $userInterval : $Interval;
$subsys= ($userSubsys ne '') ? $userSubsys : $SubsysCore;
error('invalid value for --lustopts') if $lustOpts ne '' && $lustOpts!~/^[BDMOR]+$/;
error('invalid value for --nfsopts') if $nfsOpts ne '' && $nfsOpts ne 'z';
error('invalid value for --memopts') if $memOpts ne '' && $memOpts!~/^[pPsRV]+$/;
error('--memopts R cannot be user with any of [psPV]') if $memOpts=~/R/ && $memOpts=~/[psPV]/;
error("--tcpfilt only applies to -st or -sT") if $tcpFilt ne '' && $subsys!~/t/i;
error("only valid --tcpopts values are 'cituIT'") if $tcpFilt ne '' && $tcpFilt!~/^[cituIT]+$/;
$tcpFilt=$tcpFiltDefault if $tcpFilt eq '' && $playback eq '';
# NOTE - technically we could allow fractional polling intervals without
# HiRes, but then we couldn't properly report the times.
if ($interval=~/\./ && !$hiResFlag)
{
$interval=int($interval+.5);
$interval=1 if $interval==0;
print "need to install HiRes to use fractional intervals, so rounding to $interval\n";
}
# ultimately we only use when doing process data
error("password file '$passwdFile' doesn't exist") if $passwdFile ne '' && !-e $passwdFile;
$passwdFile=$PasswdFile if $passwdFile eq '';
# S u b s y s / I n t e r v a l R e s o l u t i o n
# This needs to get done as soon a possible...
# Set default interval and subsystems for interactive mode unless already
# set, noting the default values above are for daemon mode. To be consistent,
# we also need to reset $Interval and $SubsysDef noting if one sets a
# secondary interval but not the primary, we need to prepend it with 1 and
# keep the secondary
if (!$daemonFlag)
{
$interval=$Interval=1 if $userInterval eq '' && !$showColFlag;
if ($showColFlag)
{
error('-c conflicts with --showcolheaders') if $count!=-1;
error('-i conflicts with --showcolheaders') if $userInterval ne '';
$interval=0;
$interval='0:0' if $subsys=~/[YZ]/;
$interval='0:0:0' if $subsys=~/E/;
$quietFlag=1; # suppress 'waiting...' startup message
}
if ($userInterval ne '' && $userInterval=~/^(:.*)/)
{
$interval="1$userInterval";
$Interval=1;
}
$SubsysDef=$SubsysDefInt;
$subsys=$SubsysDef if $userSubsys eq '';
}
# subsystems - must preceed +
# special option -s-all disables ALL subsystems which is basically the only way to
# disable all subsystems when you want to play back one or more explicit imports
# so we need to to allow if the ONLY thing that follows -s
error("+/- must start -s arguments if used") if $subsys=~/[+-]/ && $subsys!~/^[+-]/;
error("-s-all only allowed with -p") if $subsys eq '-all' && $playback eq '';
error("invalid subsystem '$subsys'") if $userSubsys ne '-all' && $subsys!~/^[-+$SubsysCore$SubsysExcore$SubsysDet]+$/;
$subsys=mergeSubsys($SubsysDef);
# note that -p, --procanalyze, --slabanalyze and --top can change $subsys
# also be sure to note if the user typed --verbose
$userVerbose=$verboseFlag;
setOutputFormat();
# switch validations once we know whether brief or verbose
error("only choose 1 of -oA and --stats") if $statsFlag>1;
error("statistics not allowed in verbose mode") if $statsFlag && $verboseFlag;
error("statistics not allowed interactively") if $statsFlag && $playback eq '';
error("--statopts required --stats") if $statOpts ne '' && !$statsFlag;
error("valid --statopts are [ais]") if $statOpts ne '' && $statOpts!~/[ais]/;
$headerRepeat=0 if $statsFlag && $statOpts!~/i/; # force single header line when not including interval data
# S p e c i a l F o r m a t s
if ($procAnalFlag || $slabAnalFlag)
{
error("--procanalyze/--slabanalyze require -p") if $playback eq '';
error("--procanalyze/--slabanalyze require -f") if $filename eq '';
error("--procanalyze/--slabanalyze do not support --utc") if $utcFlag;
error("--procanalyze/--slabanalyze with -P requires -s") if $plotFlag && $userSubsys eq '';
error("sorry, but no + or - with -s and analyze mode") if $userSubsys=~/[+-]/;
# No default from playback file in this mode, so go by whatever user
# specificed with -s and if no Y/Z, stick one in there and then make
# user $userSubsys and $subsys agree so initFormat() won't diddle
# the values.
$slabAnalOnlyFlag=($slabAnalFlag && $userSubsys!~/Y/) ? 1 : 0;
$procAnalOnlyFlag=($procAnalFlag && $userSubsys!~/Z/) ? 1 : 0;
$userSubsys.='Y' if $slabAnalOnlyFlag;
$userSubsys.='Z' if $procAnalOnlyFlag;
$subsys=$userSubsys;
$plotFlag=1;
}
# We have to wait for '$subsys' to be defined before handling top and it
# felt right to keep the code together with --procanalyze/--slabanalyze.
# --top forces $homeFlag if not in playback mode. if no process interval
# specified set it to the monitoring one.
$temp=$SubsysDet;
$temp=~s/YZ//;
$detailFlag=($subsys=~/[$temp]/) ? 1 : 0;
if ($topOpts ne '')
{
# Don't diddle original setting in '$userSubsys', use a copy!
# Subtle - the verbose flag wouldn't have been set if ONLY processes or slabs and
# it should be. Similarly, Y/Z should not be considered when looking to see is
# same columns in verbose mode.
my $tempSubsys=$userSubsys;
$tempSubsys=~s/[YZ]//g;
$verboseFlag=1 if $tempSubsys eq '';
$sameColsFlag=1 if $verboseFlag && length($tempSubsys)==1;
$briefFlag=($verboseFlag) ? 0 : 1;
my $subsysSize=0;
if ($tempSubsys ne '' && $playback eq '')
{
if (!$verboseFlag || $sameColsFlag)
{
# in brief or single-subsys verbose mode the area size if fixed by --hr
$subsysSize=$headerRepeat+2;
}
else
{
# multi-subsys verbose mode is driven by the number of subsystems but if
# there are any details, it's up to the users choice of --hr
$subsysSize=length($tempSubsys)*3;
$subsysSize++ if $tempSubsys=~/m/;
$subsysSize=$headerRepeat if $detailFlag;
}
$scrollEnd=$subsysSize+1;
}
($topType, $numTop)=split(/,/, $topOpts);
$topType='time' if $topType eq '';
# enough of these to warrant setting a flag
$topIOFlag=($topType=~/io|kb|sys$|cncl/) ? 1 : 0;
$termHeight=12 if $playback ne '';
$numTop=$termHeight-$scrollEnd-2 if !defined($numTop) || $numTop==-1;
#print "HEIGHT: $termHeight SUBSIZE: $subsysSize HR: $headerRepeat NUMTOP: $numTop\n";
$topProcFlag=(defined($TopProcTypes{$topType})) ? 1 : 0;
$topSlabFlag=(defined($TopSlabTypes{$topType})) ? 1 : 0;
error("not enough lines in window for display")
if $numTop<1;
error("invalid --top type. see --showtopopts for list")
if $topProcFlag==0 && $topSlabFlag==0;
error("you cannot select process and slab subsystems in --top mode")
if ($subsys=~/Y/ && $subsys=~/Z/) ||
($subsys=~/Y/ && $topProcFlag) || ($subsys=~/Z/ && $topSlabFlag);
# if sorting by v/n context switches, force --procopts x if not specified
$procOpts.='x' if $topType=~/vctx|nctx/ && $procOpts!~/x/;
if ($playback eq '')
{
$homeFlag=1;
$subsys=(defined($TopProcTypes{$topType})) ? "${tempSubsys}Z" : "${tempSubsys}Y";
$interval.=":$interval" if $interval!~/:/;
}
}
# I m p o r t
if ($import ne '')
{
# Default mode for --import is NO user defined subsystem in interactive mode.
# All must be explicitly defined
$subsys='' if !$daemonFlag && $userSubsys eq '';
foreach my $imp (split(/:/, $import))
{
$impString=$imp;
$impDetFlag[$impNumMods]=0;
$impNumMods++;
# The following chunks based somewhat on --export code, except OPTS is a string
($impName, $impOpts)=split(/,/, $impString, 2);
$impName.=".ph" if $impName!~/\./;
# If the import file itself doesn't exist in current directory, try $ReqDir
my $tempName=$impName;
$impName="$ReqDir/$impName" if !-e $impName;
if (!-e "$impName")
{
my $temp="can't find import file '$tempName' in ./";
$temp.=" OR $ReqDir/" if $ReqDir ne '.';
error($temp) if !-e "$impName";
}
require $impName;
# the basename is the name of the function and also remove extension.
$impName=basename($impName);
$impName=(split(/\./, $impName))[0];
push @impOpts, $impOpts;
push @impInit, "${impName}Init";
push @impGetData, "${impName}GetData";
push @impGetHeader, "${impName}GetHeader";
push @impInitInterval, "${impName}InitInterval";
push @impIntervalEnd, "${impName}IntervalEnd";
push @impAnalyze, "${impName}Analyze";
push @impUpdateHeader, "${impName}UpdateHeader";
push @impPrintBrief, "${impName}PrintBrief";
push @impPrintVerbose, "${impName}PrintVerbose";
push @impPrintPlot, "${impName}PrintPlot";
push @impPrintExport, "${impName}PrintExport";
}
# Call REQUIRED initialization routines in reverse so if we have to
# delete anything we won't have to deal with overlap
$impSummaryFlag=$impDetailFlag=0;
for (my $i=($impNumMods-1); $i>=0; $i--)
{
my $status=&{$impInit[$i]}(\$impOpts[$i], \$impKey[$i]);
if ($status==-1)
{
splice(@impOpts, $i, 1);
splice(@impKey, $i, 1);
splice(@impInit, $i, 1);
splice(@impGetData, $i, 1);
splice(@impGetHeader, $i, 1);
splice(@impInitInterval, $i, 1);
splice(@impIntervalEnd, $i, 1);
splice(@impAnalyze, $i, 1);
splice(@impUpdateHeader, $i, 1);
splice(@impPrintBrief, $i, 1);
splice(@impPrintVerbose, $i, 1);
splice(@impPrintPlot, $i, 1);
splice(@impPrintExport, $i, 1);
$impNumMods--;
next;
}
# We need to know if any module has summary or data in case one one else does
# and we're in plot format so newlog() will know to open tab file. This also
# helps optimize some of the print routines.
$impSummaryFlag++ if $impOpts[$i]=~/s/;
$impDetailFlag++ if $impOpts[$i]=~/d/;
}
# Reset output formatting based on the modules we just loaded
print "Reset output flags\n" if $debug & 1;
setOutputFormat();
}
# E x p o r t M o d u l e s
# since we might want to diddle with things like $subsys or fake out other
# switches, we need to load/initialize things early. We may also need a
# call to a pre-execution init module later...
# This one needs more explanation. Most export modules expect to either log
# to a file or send their output over a socket and so collectl will generate
# an error if you try to do so without -f or -A. BUT modules like ganglia
# or graphite do their own communications and so need to set this flag to
# defeat that message in case they don't want to locally log data.
$exportComm=0;
if ($export ne '')
{
# By design, if you specify --export and -f and have a socket open, the exported
# data goes over the socket and we write either a raw or plot file to the dir
# pointed to by -f. If not -P, we always write a raw file
$rawtooFlag=1 if $sockFlag && $filename ne '' && !$plotFlag;
$verboseFlag=1;
($expName, @expOpts)=split(/,/, $export);
$expName.=".ph" if $expName!~/\./;
# If the export file itself doesn't exist in current directory, try $ReqDir
my $tempName=$expName;
$expName="$ReqDir/$expName" if !-e $expName;
if (!-e "$expName")
{
my $temp="can't find export file '$tempName' in ./";
$temp.=" OR $ReqDir/" if $ReqDir ne '.';
error($temp);
}
require $expName;
# the basename is the name of the function and also remove extension.
$expName=basename($expName);
$expName=(split(/\./, $expName))[0];
}
# S i m p l e S w i t c h C h e c k s
$utcFlag=1 if $options=~/U/;
# should I migrate a lot of other simple tests here?
error("you cannot specify -f with --top") if $topOpts ne '' && $filename ne '';
error("--home does not apply to -p") if $homeFlag && $playback ne '';
error("--envopts M does not apply to -P") if $userEnvOpts ne '' && $userEnvOpts=~/M/ && $plotFlag;
error("--envopts are only fptCFMT and/or a number") if $userEnvOpts ne '' && $userEnvOpts!~/^[fptCFMT0-9]+$/;
error("--envrules does not exist") if $envRules ne '' && !-e $envRules;
error("--grep only applies to -p") if $grepPattern ne '' && $playback eq '';
error('--headerrepeat must be an integer') if $headerRepeat!~/^[\-]?\d+$/;
error('--headerrepeat must be >= -1') if $headerRepeat<-1;
error("-i not allowed with -p") if $userInterval ne '' && $playback ne '';
error("--rawtoo does not work in playback mode") if $rawtooFlag && $playback ne '';
error("--rawtoo requires -f") if $rawtooFlag && $filename eq '';
error("--rawtoo requires -P or --export") if $rawtooFlag && !$plotFlag && $export eq '';
error("--rawtoo and -P requires -f") if $rawtooFlag && $plotFlag && $filename eq '';
error("--rawtoo cannot be used with -p") if $rawtooFlag && $playback ne '';
error("-ou/--utc only apply to -P format") if $utcFlag && !$plotFlag;
error("can't mix UTC time with other time formats") if $utcFlag && $options=~/[dDT]/;
error("-oz only applies to -P files") if $options=~/z/ && !$plotFlag;
error("--sep cannot be a '%'") if defined($SEP) && $SEP eq '%';
error("--sep only applies to plot format") if defined($SEP) && !$plotFlag;
error("--sep much be 1 character or a number") if defined($SEP) && length($SEP)>1 && $SEP!~/^\d+$/;
error('--showheader not allowed with -f') if $filename ne '' && $showHeaderFlag;
error("--showheader in collection mode only supported on linux")
if $PcFlag && $playback eq '' && $showHeaderFlag;
error('--showmergedheader not allowed with -f') if $filename ne '' && $showMergedFlag;
error('--showcolheaders not allowed with -f') if $filename ne '' && $showColFlag;
error('--showcolheaders -sE can only be run by root') if $showColFlag && $subsys=~/E/ && !$rootFlag;
error("--align require HiRes time module") if $alignFlag && !$hiResFlag;
error('--umask can only be set by root') if $umask ne '' && !$rootFlag;
error('-sT can only be used with -f or -P') if $subsys=~/T/ && !$plotFlag && $filename eq '';
# if user enters --envOpts
if ($userEnvOpts ne '')
{
# remove ALL ipmi data types if user specified any, then add in ALL user options
# which could include formatting options
$envOpts=~s/[fpt]+//g if $userEnvOpts=~/[fpt]/;
$envOpts.=$userEnvOpts;
}
$allThreadFlag=($procOpts=~/t/) ? 1 : 0;
# The separator is either a space if not defined or the character supplied if
# non-numeric. If it is numeric assume decimal and convert to the associated
# char code (eg 9=tab).
$SEP=' ' if !defined($SEP);
$SEP=sprintf("%c", $SEP) if $SEP=~/\d+/;
# Even though users warning about this in docs, it's too easy to forget
# at least for me, so make sure no quotes in filters
$netFilt=~s/['"]//g;
$dskFilt=~s/['"]//g;
$rawNetFilter=~s/['"]//g;
$rawDskFilter=~s/['"]//g;
# Both kinds of DISK and NETWORK filtering
# Remember, this filter overrides the one in collectl.conf
if ($rawDskFilter ne '')
{
$DiskFilter=$rawDskFilter;
$DiskFilterFlag=1;
}
# This is applied AFTER the raw disk records are read and possibly filtered
$dskFiltKeep='';
$dskFiltIgnore='';
my $ignoreFlag=($dskFilt=~s/^\^//) ? 1 : 0;
foreach my $disk (split(/,/, $dskFilt))
{
$dskFiltIgnore.="|$disk" if $ignoreFlag;
$dskFiltKeep.= "|$disk" if !$ignoreFlag;
}
$dskFiltKeep=~s/^\|//;
$dskFiltIgnore=~s/^\|//;
print "DskFilt - Ignore: $dskFiltIgnore Keep: $dskFiltKeep\n" if $debug & 1;
# Unlike the raw disk filter which uses a flag to decided whether or not to use
# if, if the raw net filter is non-blank its very presence is the flag so nothing
# to set
$netFiltKeep='';
$netFiltIgnore='';
$ignoreFlag=($netFilt=~s/^\^//) ? 1 : 0;
foreach my $net (split(/,/, $netFilt))
{
$netFiltIgnore.="|$net" if $ignoreFlag;
$netFiltKeep.= "|$net" if !$ignoreFlag;
}
$netFiltKeep=~s/^\|//;
$netFiltIgnore=~s/^\|//;
print "NetFilt - Ignore: $netFiltIgnore Keep: $netFiltKeep\n" if $debug & 1;
# raw net/dsk filters are different in that they're applied at data collection time
# and so will never even make it to the raw file. for ease of use, one can separate
# multiple entries by commas or pipe, but to make them work in the regx, convert the
# commas to pipes
$rawDskFilter=~s/,/|/g;
$rawDskIgnore=~s/,/|/g;
$rawNetFilter=~s/,/|/g;
$rawNetIgnore=~s/,/|/g;
print "RawDsk - Ignore: $rawDskIgnore Keep: $rawDskFilter\n" if $debug & 1;
print "RawNet - Ignore: $rawNetIgnore Keep: $rawNetFilter\n" if $debug & 1;
error("--dskopts f only applies to -sD") if $dskOpts=~/f/ && $subsys!~/D/;
error("--dskopts z only applies to -sD") if $dskOpts=~/z/ && $subsys!~/D/;
error("only valid value for --cpuopts is 'z'") if $cpuOpts ne '' && $cpuOpts!~/^[z]+$/;
error("only valid values for --dskopts are 'fioz'") if $dskOpts ne '' && $dskOpts!~/^[fioz]+$/;
error("only valid value for --xopts is 'i'") if $xOpts ne '' && $xOpts!~/^[i]+$/;
$netOptsW=5; # minumum width
if ($netOpts ne '')
{
error("--netopts only applies to -sn or -sN") if $subsys!~/n/i;
error("only valid --netopts values are 'eEiow'") if $netOpts ne '' && $netOpts!~/^[eEiow0-9]+$/;
if ($netOpts=~/w/)
{
error("--netopts -w only applies to -sN") if $subsys!~/N/;
error("--netopts w must be followed by width") if $netOpts!~/w(\d+)/;
$netOptsW=$1;
error("--netopts width must be at least 5") if $netOptsW<5;
}
}
# This is applied AFTER the interrupt records are read and possibly filtered
$intFiltKeep='';
$intFiltIgnore='';
$ignoreFlag=($intFilt=~s/^\^//) ? 1 : 0;
foreach my $int (split(/,/, $intFilt))
{
$intFiltIgnore.="|$int" if $ignoreFlag;
$intFiltKeep.= "|$int" if !$ignoreFlag;
}
$intFiltKeep=~s/^\|//;
$intFiltIgnore=~s/^\|//;
print "IntFilt - Ignore: $intFiltIgnore Keep: $intFiltKeep\n" if $debug & 1 && $intFilt ne '';
# L i n u x S p e c i f i c
if (!$PcFlag)
{
# This matches THIS host, but in playback mode will be reset to the target
$Kernel=`uname -r`;
chomp $Kernel;
error("collectl no longer supports 2.4 kernels") if $Kernel=~/^2\.4/;
$LocalTimeZone=`date +%z`;
chomp $LocalTimeZone;
# Some distros put lspci in /usr/sbin and others in /usr/bin, so take one last look in
# those before complaining, but only if in record mode AND only if looking at interconnects
if (!-e $Lspci && $playback eq '' && $subsys=~/x/i)
{
$Lspci=(-e '/usr/sbin/lspci') ? '/usr/sbin/lspci' : '/usr/bin/lspci';
if (!-e "/usr/sbin/lspci" && !-e "/usr/bin/lspci")
{
pushmsg('W', "-sx disabled because 'lspci' not in $Lspci or '/usr/sbin' or '/usr/bin'");
pushmsg('W', "If somewhere else, move it or define in collectl.conf");
$xFlag=$XFlag=0;
$subsys=~s/x//ig;
}
}
if (!-e $Dmidecode && $playback eq '')
{
# we really only care about the message is doing -sE
pushmsg('W', "cannot find '$Dmidecode' so can't determine hardware Product Name") if $subsys=~/E/;
$Dmidecode='';
$ProductName='Unknown';
}
# Set protections for output files
umask oct($umask) or error("Couldn't set umask to $umask") if $umask ne '' && $rootFlag;
}
# C o m m o n I n i t i a l i z a t i o n
# We always want to flush terminal buffer in case we're using pipes.
$|=1;
# We need to know where we're logging to so set a couple of flags
$logToFileFlag=0;
$rawFlag=$rawtooFlag;
if ($filename ne '')
{
$rawFlag=1 if !$plotFlag && $export eq '';
$logToFileFlag=1 if $rawFlag || $plotFlag;
}
printf "RawFlag: %d PlotFlag: %d Repeat: %d Log2Flag: %d Export: %s\n",
$rawFlag, $plotFlag, $headerRepeat, $logToFileFlag, $export if $debug & 1;
($lustreSvcs, $lustreConfigInt)=split(/:/, $lustreSvcs);
$lustreSvcs="" if !defined($lustreSvcs);
$lustreConfigInt=$LustreConfigInt if !defined($lustreConfigInt);
error("Valid values for --lustsvcs are any combinations of cmoCMO")
if $lustreSvcs!~/^[cmo]*$/i;
error("lustre config check interval must be numeric")
if $lustreConfigInt!~/^\d+$/;
# some restrictions of plot format -- can't send to terminal for slabs or
# processes unless only 1 subsystem selected. quite frankly I see no reason
# to ever do it but there are so damn many other odd switch combos we might
# as well catch these too.
error("to display on terminal using -sY with -P requires only -sY")
if $plotFlag && $filename eq '' && $subsys=~/Y/ && length($subsys)>1;
error("to display on terminal using -sZ with -P requires only -sZ")
if $plotFlag && $filename eq '' && $subsys=~/Z/ && length($subsys)>1;
# No great place to put this, but at least here it's in you face! There are times
# when someone may want to automate the running of collectl to playback/convert
# logs from crontab for the day before and this is the easiest way to do that.
# While we're at it, there may be some other 'early' checks that need to be make
# in playback mode.
if ($playback ne "")
{
($day, $mon, $year)=(localtime(time))[3..5];
$today=sprintf("%d%02d%02d", $year+1900, $mon+1, $day);
$playback=~s/TODAY/$today/;
($day, $mon, $year)=(localtime(time-86400))[3..5];
$yesterday=sprintf("%d%02d%02d", $year+1900, $mon+1, $day);
$playback=~s/YESTERDAY/$yesterday/;
error("sorry, but --procfilt not allowed in -p mode. consider grep")
if $procFilt ne '';
error("sorry, but --slabfilt not allowed in -p mode. consider grep")
if $slabFilt ne '';
}
# linux box?
if ($SrcArch!~/linux/)
{
error("record mode only runs on linux") if $playback eq "";
error("-N only works on linux") if $niceFlag;
}
# daemon
if ($daemonFlag)
{
error("no debugging allowed with -D") if $debug;
error("-D requires -f OR -A server") if $filename eq '' && !$serverFlag && !$exportComm;
error("-p not allowed with -D") if $playback ne "";
if (-e $PidFile)
{
# see if this pid matches a version of collectl. If not, we'll overwrite
# it further on so not to worry, but at least record a warning.
$pid=`$Cat $PidFile`;
chomp $pid;
@ps=`ps axo pid,command`;
foreach my $line (@ps)
{
$line=~s/^\s+//; # trim leading whitespace for short pids
($procPid, $procCommand)=split(/\s+/, $line, 2);
if ($procPid eq $pid && $procCommand=~/collectl/)
{
error("a daemonized collectl already running");
}
}
}
}
# count
if ($count!=-1)
{
error("-c must be numeric") if $count!~/^\d+$/;
error("-c conflicts with -r and -R") if $rollLog ne "" || $runTime ne "";
$count++ # since we actually need 1 extra interval
}
if ($limits ne '')
{
error("-l only makes sense for -s D/L/l") if $subsys!~/[DLl]/;
@limits=split(/-/, $limits);
foreach $limit (@limits)
{
error("invalid value for -l: $limit")
if $limit!~/^SVC:|^IOS:|^LusKBS:|^LusReints:|^OR|^AND/;
($name,$value)=split(/:/, $limit);
$limBool=0 if $name=~/OR/;
$limBool=1 if $name=~/AND/;
next if $name=~/AND|OR/;
error("-l SVC and IOS only apply to -sD") if $name!~/^Lus/ && $subsys=~/L/;
error("-l LusKBS and LusReint only apply to -sL") if $name=~/^Lus/ && $subsys=~/D/;
error("limit for $limit not numeric") if $value!~/^\d+$/;
$limSVC=$value if $name=~/SVC/;
$limIOS=$value if $name=~/IOS/;
$limLusKBS=$value if $name=~/LusKBS/;
$limLusReints=$value if $name=~/LusReints/;
}
}
# options
error("invalid option") if $options ne "" && $options!~/^[\^12acdDGgimnTuUxXz]+$/g;
error("-oi only supported interactively with -P to terminal")
if $options=~/i/ && ($playback ne '' || !$plotFlag || $filename ne '');
$miniDateFlag=($options=~/d/i) ? 1 : 0;
$miniTimeFlag=($options=~/T/) ? 1 : 0;
error("use only 1 of -o dDT")
if ($miniDateFlag && $miniTimeFlag) || ($options=~/d/ && $options=~/D/);
error("--home only applies to terminal output")
if $homeFlag && $filename ne "";
error("--home cannot be used with -A")
if $homeFlag && $sockFlag;
error("option $1 only apply to -P")
if !$plotFlag && $options=~/([12ac])/;
error("-oa conflicts with -oc")
if $options=~/a/ && $options=~/c/;
error("-oa conflicts with -ou")
if $options=~/a/ && $options=~/u/;
if (!$hiResFlag && $options=~/m/)
{
print "need to install HiRes to report fractional time with -om, so ignoring\n";
$options=~s/m//;
}
$pidOnlyFlag=($procOpts=~/p/) ? 1 : 0;
# We always compress files unless zlib not there or explicity turned off
$zFlag=($options=~/z/ || $filename eq "") ? 0 : 1;
if (!$zlibFlag && $zFlag)
{
$options.="z";
$zFlag=0;
pushmsg("W", "Zlib not installed so can't compress raw file(s). Use --quiet to disable this warning.") if $rawFlag;
pushmsg("W", "Zlib not installed so can't compress plot file(s). Use -oz to get rid of this warning.") if $plotFlag;
}
$precision=($options=~/(\d+)/) ? $1 : 0;
$FS=".${precision}f";
# playback mode specific
error('--showmerged only applies to playback mode') if $playback eq '' && $showMergedFlag;
error('--extract only applies to playback mode') if $playback eq '' && $extract ne '';
if ($playback ne "")
{
error("-p not allowed with -F") if $flush ne '';
error("--offsettime must be in seconds with optional leading '-'")
if defined($offsetTime) && $offsetTime!~/^-?\d+/;
$playback=~s/['"]//g; # in case quotes passed through from script
$playback=~s/,/ /g; # so glob below will work
error("--align only applies to record mode") if $alignFlag;
error("-p filename must end in '*', 'raw' or 'gz'")
if $playback!~/\*$|raw$|gz$/;
error("MUST specify -P if -p and -f") if $filename ne "" and !$plotFlag;
if ($extract ne '')
{
$extractMode=1;
error("-s not allowed in 'extract' mode") if $userSubsys ne '';
error("--from OR --thru required in 'extract' mode") if !defined($from) && !defined($thru);
}
# Quick check to make sure at least one file matches playback string
my $foundFlag=0;
foreach $file (glob($playback))
{
$foundFlag=1;
# this is a great place to print headers since we're already looping through glob
if ($showHeaderFlag)
{
next if $file!~/raw/;
# remember, this has to work on a pc as well, so can't use linux commands
print "$file\n";
my $return;
$return=open TMP, "<$file" if $file!~/gz$/;
$return=($ZTMP=Compress::Zlib::gzopen($file, 'rb')) if $file=~/gz$/;
logmsg("F", "Couldn't open '$file' for reading") if !defined($return) || $return<1;
while (1)
{
$line= if $file!~/gz$/;
$ZTMP->gzreadline($line) if $file=~/gz$/;
last if $line!~/^#/;
print $line;
}
print "\n";
close TMP if $file!~/gz$/;
$ZTMP->gzclose() if $file=~/gz$/;
}
}
error("can't find any files matching '$playback'") if !$foundFlag;
exit(0) if $showHeaderFlag;
}
# end time
$purgeDays=0;
if (defined($from) || defined($thru))
{
error("--from/--thru only apply to -p")
if $playback eq '';
error("do not specify 2 times with --thru")
if defined($thru) && index($thru, '-')!=-1;
error("do not specify 2 times with --from and also use --thru")
if defined($from) && defined($thru) && index($from, '-')!=-1;
# Parse switches and handle those that only specify a date
($from, $thru)=split(/-/, $from) if !defined($thru);
($fromDate,$fromTime)=checkTime('--from', $from) if defined($from);
($thruDate,$thruTime)=checkTime('--thru', $thru) if defined($thru);
}
$fromDate=0 if !defined($fromDate); # 0 means all dates
$thruDate=0 if !defined($thruDate);
$fromTime='000000' if !defined($fromTime);
$thruTime='235959' if !defined($thruTime);
print "From: $fromDate $fromTime Thru: $thruDate $thruTime\n" if $debug & 1;
$endSecs=0;
if ($runTime ne "")
{
error("pick either -r or -R") if $rollLog ne "";
error("invalid -R format") if $runTime!~/^(\d+)[wdhms]{1}$/;
$endSecs=$1;
$endSecs*=60 if $runTime=~/m/;
$endSecs*=3600 if $runTime=~/h/;
$endSecs*=86400 if $runTime=~/d/;
$endSecs*=604800 if $runTime=~/w/;
$endSecs+=time;
}
# log file rollover
my $rollSecs=0;
my $expectedHour;
if ($rollLog ne '')
{
error("-r requires -f") if $filename eq "";
($rollTime,$purgeDays,$rollIncr)=split(/,/, $rollLog);
($purgeDays, $purgeMons)=split(/:/, $purgeDays);
$rollIncr=60*24 if !defined($rollIncr) || $rollIncr eq '';
# default is 7 days for data and 12 months for logs
$purgeDays=7 if !defined($purgeDays) || $purgeDays eq '';
$purgeMons=12 if !defined($purgeMons);
error("-r time must be in HH:MM format") if $rollTime!~/^\d{2}:\d{2}$/;
($rollHour, $rollMin)=split(/:/, $rollTime);
error("-r purge days must be an integer") if $purgeDays!~/^\d+$/;
error("-r purge months must be an integer") if $purgeMons!~/^\d+$/;
error("-r increment must be an integer") if $rollIncr!~/^\d+$/;
error("-r time invalid") if $rollHour>23 || $rollMin>59;
error("-r increment must be a factor of 24 hours")
if int(1440/$rollIncr)*$rollIncr!=1440;
error("if -r increment>1 hour, must be multiple of 1 hour")
if $rollIncr>60 && int($rollIncr/60)*60!=$rollIncr;
error("roll time must be specified in 1st interval")
if ($rollHour*60+$rollMin)>$rollIncr;
# Getting the time to the next interval can be tricky because we have to
# worry about daylight savings time. This IS further complicated by
# having to deal with intervals. The safest thing to do is using brute-force.
# I also have to write the following down because I know I'll forget it and
# think it's a bug! Assume you're going to roll every two hours (or more) and it's
# midnite of the day to move clocks forward (probably never going happen but...).
# 2 hours from midnight is 3AM! so we subtract an hour and now since we're before the
# time change we create a logfile with a time of 1AM. BUT the next log gets created
# at AM and everyone is happy!
# We start at the first interval of the day and then step forward until we
# pass our current time. Then we see if DST is involved and then we're done!
# Note however, if the interval is an hour or less, DST takes care of itself!
# Step 1 - Get current date/time
my ($sec, $min, $hour, $day, $mon, $year)=localtime(time);
my $timeNow=sprintf "%d%02d%02d %02d:%02d:%02d",
$year+1900, $mon+1, $day, $hour, $min, $sec;
$rollToday=timelocal(0, $rollMin, $rollHour, $day, $mon, $year);
# Step 2 - step through each increment (note in most cases there is only 1!)
# looking for each one > now
my $timeToRoll;
$expectedHour=$rollHour;
foreach ($rollSecs=$rollToday;; $rollSecs+=$rollIncr*60)
{
# Get the corresponding time and if not the first one see if the
# time was changed
my ($sec, $min, $hour, $day, $mon, $year)=localtime($rollSecs);
$timeToRoll=sprintf "%d%02d%02d %02d:%02d:%02d", $year+1900, $mon+1, $day, $hour, $min, $sec;
#print "CurTime: $timeToRoll CurHour: $hour ExpectedHour: $expectedHour\n";
if ($rollIncr>60)
{
# Tricky... We can have expected hour differ from the current one by
# exactly 1 hour when we hit a DST time change. However, while a
# simple subtraction will yield +/- 1, the one special case is when
# we're rolling logs at 00:00 and get an hour of 23, which generates a
# diff of -23 when we really want +1.
my $diff=($expectedHour-$hour);
$specialFlag=($diff==-23) ? 1 : 0;
$diff=1 if $specialFlag;
$rollSecs+=$diff*3600; # diff is USUALLY 0
# When in this 'special' situation, '$timeToRoll' is pointing to the previous
# day so we need to reset $timeToRoll, but only AFTER we updated rollSecs.
if ($specialFlag)
{
($sec, $min, $hour, $day, $mon, $year)=localtime($rollSecs);
$timeToRoll=sprintf "%d%02d%02d %02d:%02d:%02d", $year+1900, $mon+1, $day, $hour, $min, $sec;
}
$expectedHour+=$rollIncr/60;
$expectedHour%=24;
}
last if $timeToRoll gt $timeNow;
}
($sec, $min, $hour, $day, $mon, $year)=localtime($rollSecs);
$rollFirst=sprintf "%d%02d%02d %02d:%02d:%02d", $year+1900, $mon+1, $day, $hour, $min, $sec;
pushmsg("I", "First log rollover will be: $rollFirst");
}
# for --home we do some vt100 cursor control
if ($homeFlag)
{
$home=sprintf("%c[H", 27); # top of display
$clr=sprintf("%c[J", 27); # clear to end of display
$clscr="$home$clr"; # clear screen
$cleol=sprintf("%c[K", 27); # clear to end of line
}
# if -N, set priority to 20
`renice 20 $$` if $niceFlag;
# Couldn't find anywhere else to put this one...
error("-sT only works with -P for now (too much data)")
if $TFlag && !$plotFlag;
# get parent pid so we can check later to see it still there
$stat=`cat /proc/$$/stat`;
$myPpid=(split(/\s+/, $stat))[3];
###############################
# P l a y b a c k M o d e
###############################
if ($playback ne '')
{
# Select all files that need to be processed based on dates
my $numSelected=0;
my ($firstFileDate, $lastFileDate, $lastPrefix)=(0,0,'');
my $pushed='';
while (my $file=glob($playback))
{
next if $file!~/(.*)-(\d{8})-(\d{6})\.(raw[p]*)/;
my $prefix= $1;
my $fileDate=$2;
my $fileTime=$3;
# we never look at rawp files when doing stats
next if $file=~/rawp/ && $statsFlag;
$firstFileDate=$fileDate if $firstFileDate==0;
$numSelected++;
# F i l t e r O u t F i l e s N e w e r T h a n t h r u D a t e
# If there IS a thru date, ignore any files start were beyond it.
next if ($thruDate ne 0) && (($fileDate > $thruDate) || (($fileDate == $thruDate) && ($fileTime > $thruTime)));
# and finally, if no from OR thru dates, the thru time is applied against all files
# so skip any files created after the from time
next if ($fromDate eq 0) && ($thruDate eq 0) && ($fileTime>$thruTime);
# New functionality for V3.5.1: only apply other filters if wildcards NOT in filename
# since files CAN contain data beyond their date stamp.
if ($playback!~/\*/)
{
push @playbackList, $file;
next;
}
# A p p l y F i l t e r s T o W i l d c a r d e d F i l e n a m e s
# We only get here is a wildcard in the file list. If it's from date is early than specified
# ignore it, remembering if it did have data that crossed midnight we'll never know. Those
# MUST be processes w/o wild cards in their names. If this ever becomes an issue we could always
# look inside the header here, but that's more work than currently deemed worth it.
next if ($fromDate ne 0) && ($fileDate < $fromDate);
# This is the magic AND there are 3 cases all of which have the common test of
# there needs to be a file with a different basename (in case we're doing a rawp
# file and there is alreay a raw file there) on the stack and the current file
# is < $fromTime, in which case we might NOT want to process the file(s) on the
# top of the stack under the the following cases:
# - if we have a from date this only applies to files for the from date
# - if no from date but a thru date, apply time to files of first date
# - if no from AND thru dates this test applies to ALL dates but only
# pop files for the same date which are by definition too 'young'
if ($file!~/$pushed/ && ($lastFileDate!=0) && ($fileTime<$fromTime) && ($prefix eq $lastPrefix))
{
if (($fromDate!=0 && $fileDate==$fromDate) ||
($fromDate==0 && $thruDate!=0 && $fileDate==$firstFileDate) ||
($fromDate==0 && $thruDate==0 && $fileDate==$lastFileDate))
{
my $popped=pop(@playbackList);
$popped=quotemeta((split(/\./, $popped))[0]);
# get rid of companion file if there is one.
pop(@playbackList) if scalar(@playbackList) && $playbackList[0]=~/$popped/;
}
}
push @playbackList, $file;
$pushed=quotemeta((split(/\./, $file))[0]); # filename less extension, ready for regx
$lastPrefix=$prefix;
$lastFileDate=$fileDate;
}
$numProcessed=0;
$elapsedSecs=0;
preprocessPlayback(\@playbackList);
$doneFlag=0;
$lastPrefix=$lastHost=$prefixPrinted=$lastSubsys='';
foreach $file (@playbackList)
{
# Unfortunately we need a more unique global name for the file we're doing
$playbackFile=$file;
$rawPFlag=($file=~/\.rawp/) ? 1 : 0;
# For now, we're going to skip files in error and process the rest.
# Some day we may just want to exit on errors (or have another switch!)
$ignoreFlag=0;
foreach $key (keys %preprocErrors)
{
# some are file names and some just prefixes.
if ($file=~/$key/)
{
($type, $text)=split(/:/, $preprocErrors{$key}, 2);
$modifier=($type eq 'E') ? 'due to error:' : 'because';
logmsg($type, "*** Skipping '$file' $modifier $text ***");
$ignoreFlag=1;
next;
}
}
next if $ignoreFlag;
print "\nPlaying back $file\n" if $msgFlag || $debug & 1;
$file=~/(.*)-(\d{8})-\d{6}\.raw[p]*/;
$prefix="$1-$2";
$fileHost=$1;
$fileRoot=basename($prefix);
# if the prefix didn't change, we can't have a new host
$newPrefixFlag=$newHostFlag=0;
if ($prefix ne $lastPrefix)
{
# Remember that the prefix includes the date so the host could still be the same!
$newPrefixFlag=1;
$newHostFlag=($fileHost ne $lastHost) ? 1 : 0;
$lastHost=$fileHost;
print "NewPrefix: $newPrefixFlag NewHost: $newHostFlag\n" if $debug & 1;
undef $newSeconds[$rawPFlag] if $newHostFlag; # indicates we start anew
# For each day's set of files, we need to reset this variable so interval
# lengths are calculared correctly. Since int3 doesn't contain any rate
# data we don't care about that one.
$lastInt2Secs=0;
if ($msgFlag && defined($preprocMessages{$prefix}))
{
# Whatever the messages may be, we only want to display them once for
# each set of files, that is files with the same prefix
my $preamblePrinted=0;
for ($i=0; $i<$preprocMessages{$prefix}; $i++)
{
$key="$prefix|$i";
if ($file=~/$prefix/)
{
# messy but makes it easier on the user to only see this message when a -s change
# didn't happen because of a raw/rawp adjustment. Since changes are appended, just
# subtract first string from final one and don't report if only [YZ] remains
if ($preprocMessages{$key}=~/-s overridden/ && ($playback{$prefix}->{flags} & 1))
{
my $first=$playback{$prefix}->{subsysFirst};
my $final=$playback{$prefix}->{subsys};
$final=~s/$first//;
next if $final=~/^[YZ]*$/;
}
print " >>> Forcing configuration change(s) for '$prefix-*'\n"
if !$preamblePrinted;
print " >>> $preprocMessages{$key}\n";
$preamblePrinted=1;
}
}
}
# When we start a new prefix, that's the time to reset any variables that
# span the set of common files.
$lustreCltInfo='';
$headersPrinted=$totalCounter=$separatorCounter=0;
# Finally save the merged set of subsystems associated with all the files for
# for this prefix.
$subsysAll=$playback{$prefix}->{subsys};
}
$lastPrefix=$prefix;
print "NewPrefix: $newPrefixFlag NewHost: $newHostFlag\n" if $debug & 1;
# we need to initialize a bunch of stuff including these variables and the
# starting time for the file as well as the corresponding UTC seconds.
($recVersion, $recDate, $recTime, $recSecs, $recTZ, $recInterval, $recSubsys, $recNfsFilt, $recHeader)=initFormat($file);
error("$file was created before collectl V2.0 and so cannot be played back") if $recVersion lt '2.0';
printf "RECORDED -- Host: $Host Version: %s Date: %s Time: %s Interval: %s Subsys: $recSubsys\n",
$recVersion, $recDate, $recTime, $recInterval
if $debug & 1;
# we can't do this until we know what version of collectl recorded the file
if ($tcpFilt ne '' && $recVersion ne '' && $recVersion lt '3.6.4-1')
{
print "$file recorded with collectl V$recVersion which does not support --tcpfilt, so skipping...\n";
next;
}
$tcpFilt='T' if $subsys=~/t/i && $recVersion lt '3.6.4-1'; # only subsystem reported earlier
# Make sure at least 1 requested subsys is actually recorded OR if -s-all clear them all
# also note an empty $subsys had been set to ' ' so regx below will work. Now set it back!
$subsys='' if $userSubsys eq '-all';
my $tempSys=$subsys; # this is what we want to report
$tempSys=~s/[$recSubsys]//gi; # remove ANY that are recorded, whether summary OR detail
$subsys='' if $subsys eq ' ';
print "recSubsys: $recSubsys subsys: $subsys tempSys: $tempSys\n" if $debug & 1;
# When processing a batch of files, it's possible none of them have any of the selected subsystems,
# the best example being playing back *.gz files which have been collected with --tworaw and only
# requestion data in one typw. In those cases both files will be processed and we need to skip
# the ones w/o data. The logmsg() below only reports the message when -m included.
if (!$numProcessed && !$impNumMods && $subsys eq $tempSys)
{
logmsg("w", "none of the requested subsystems are recorded in selected file");
next;
}
loadUids($passwdFile) if $recSubsys=~/Z/;
#print "SUBSYS: $subsys RECSUBSYS: $recSubsys FLAGS: $playback{$prefix}->{flags}\n";
# if --top but user didn't specify -s too, ignore anything in header(s)
$subsys=~s/[^YZ]*//g if $topFlag && $userSubsys eq '';
$subsysAll=~s/[^YZ]*//g if $topFlag && $userSubsys eq '';
# Now that we know the subsystem it's safe to initialize a custom --export module if using one.
if ($expName ne '')
{
my $initName="${expName}Init";
&$initName(@expOpts);
}
# I wanted these 'in your face' rather than buried in 'initFormat()'.
if ($playback{$prefix}->{flags} & 1)
{
# when playing back data from BOTH files, we need to reset these if in fact something to
# print from rawp so that we'll repeat brief headers.
$headersPrinted=$totalCounter=0 if $subsys=~/[YZ]/i;
# When playing back files generated with -G and user specified -s, make sure that subsys
# only contains file-related subsystems so $subsys is consistent with the file we're processing
$subsys=~s/[YZ]//gi if $file!~/rawp/;
$subsys=~s/[^YZ]//gi if $file=~/rawp/;
next if $subsys eq ''; # in case $subsys now '' for this file
}
else
{
# no 'rawp' files associated with this prefix so if user chose 'y' in playback and no slab
# data has been recorded, ignore it so we won't put ourselves into --verbose because of it.
# NOTE - this is an exception to the rule that if the user requests a subsystem for which
# we have no data we report it as zeros.
$subsys=~s/y//gi if $recSubsys!~/y/i;
}
# the only way nfsfilt can come back null is when there is a blank nfsfilt in header
my $tempFilt=($recNfsFilt ne '' ? $recNfsFilt : 'c2,s2,c3,s3,c4,s4');
if ($nfsFilt ne '')
{
foreach my $filt (split(/,/, $nfsFilt))
{
error("'$filt' data not recorded in $file and so cannot be selected")
if $tempFilt!~/$filt/;
}
$tempFilt=$nfsFilt;
}
setNFSFlags($tempFilt);
# We can only do this test after figuring out what's in the header. NOTE that since the number
# of enabled CPUs can change dynamically when doing -sC and we've already skipped the code in
# formatit that sets the number to 0, we have to do it here too.
if ($subsys=~/j/i && $subsys!~/C/i && $plotFlag)
{
logmsg('I', "-sj or -sJ with -P also requires CPU details so adding -sC. See FAQ for details.");
$subsys.='C';
$subsysAll.='C';
$noCpusFlag=1; # we need to know elsewhere when this was done
$cpusEnabled=0 if $recSubsys=~/c/i; # if recorded, WILL be dynamically reset
}
# the way the process/slab tests work is if raw file not built with -G, look at all files.
# but IF a -G only look at rawp files.
if (($playback{$prefix}->{flags} & 1)==0 || $rawPFlag)
{
# no rawp files so these tests are pretty easy
my $skipmsg='';
$skipmsg="io" if $topIOFlag && !$processIOFlag;
$skipmsg="process" if $procAnalFlag && $recSubsys!~/Z/;
$skipmsg="slab" if $slabAnalFlag && $recSubsys!~/Y/;
if ($skipmsg ne '')
{
print " >>> Skipping file because it does not contain $skipmsg data <<<\n";
next;
}
}
# Need to reset the globals for the intervals that gets recorded in the header.
# Note the conditional on the assignments for i2 and i3. This is because they SHOULD be
# in the header as of V2.1.0 and I don't want to mask any problems if they're not.
($interval, $interval2, $interval3)=split(/:/, $recInterval);
$interval2=$Interval2 if !defined($interval2) && $recVersion lt '2.1.0';
$interval3=$Interval3 if !defined($interval2) && $recVersion lt '2.1.0';
# At this point we've initialized all the variables that will get written to the common
# header for one set of files for one day, so if the user had specified --showmerged, now
# is the best/easiest time to do it. We also need to set a flag so we only print the
# header once for each set of merged files
if ($showMergedFlag)
{
# I'm bummed I can't use '$lastPrefix', but we don't always execute the
# outer loop and can't rest it in one place common to everyone...
if ($prefix ne $prefixPrinted)
{
$commonHeader=buildCommonHeader(0);
}
$prefixPrinted=$prefix;
next;
}
# on the off chance that lustre data was collected with --lustopts but not
# played back, clear the lustre settings or else we're screw up the default
# playback mode.
$lustOpts='' if $subsys!~/l/i;
# conversely, if data was collected using --lustOpts but lustre
# wasn't active during the time this file was collected, the header will
# indicate this log does NOT contain any lustre data but the -s will and
# so we need to turn off any -lustOpts or else 'checkSubsysOpts()'
# will report a conflict.
$lustOpts=~s/B//g if $lustOpts=~/B/ && $CltFlag==0 && $OstFlag==0;
$lustOpts=~s/D//g if $lustOpts=~/D/ && $MdsFlag==0 && $OstFlag==0;
$lustOpts=~s/[MR]//g if $lustOpts=~/[MR]/ && $CltFlag==0;
# Now we can check for valid/consistent sub-options (not sure this is still
# necessary, but it shouldn't hurt). Since we can swap back and forth between
# raw and rawp, with the latter requiring verbose, always reset to the default
# of brief, unless if course user specified --verbose.
checkSubsysOpts(); # Make sure valid
$verboseFlag=1 if $userVerbose;
setOutputFormat();
# We need to set the 'coreFlag' based on whether or not any core
# subsystems will be processed.
$coreFlag=($subsys=~/[a-z]/) ? 1 : 0;
# if a specific time offset wasn't selected, find difference between
# time collectl wrote out the log and the time of the first timestamp.
if (!defined($offsetTime) && $recSecs ne '')
{
$year=substr($recDate, 0, 4);
$mon= substr($recDate, 4, 2);
$day= substr($recDate, 6, 2);
$hour=substr($recTime, 0, 2);
$min= substr($recTime, 2, 2);
$sec= substr($recTime, 4, 2);
$locSecs=timelocal($sec, $min, $hour, $day, $mon-1, $year-1900);
$timeAdjust=$locSecs-$recSecs;
}
elsif (defined($offsetTime))
{
$timeAdjust=$offsetTime; # user override of default
}
# Header already successfully read one, but what the heck...
if (!defined($recVersion))
{
logmsg("E", "Couldn't read header for $file");
next;
}
# Note - the prefix includes full path
$zInFlag=($file=~/gz$/) ? 1 : 0;
$file=~/(.*-\d{8})-\d{6}\.raw[p]*/;
$prefix=$1;
if ($prefix!~/$Host/)
{
print "ignoring $file whose header says recorded for $Host but whose name says otherwise!\n";
next;
}
# we get new output files (if writing to a file) for each prefix-date combo noting if reading
# a rawp file we might get 2, also noting that $Host is a global pointing to the current host
# being processed both in record as well as playback mode. We also need to track for terminal
# processing as well, so use a different flag for that
$key="$prefix:$recDate";
$newPrefixDate=(!defined($playback{$key})) ? 1 : 0;
if ($newPrefixDate)
{
print "Prefix: $prefix Host: $Host\n" if ($debug & 1) && !$logToFileFlag;
$headersPrintedProc=$headersPrintedSlab=$prcFileCount=0;
$newOutputFile=($filename ne '') ? 1 : 0;
$playback{$key}=1;
}
$prcFileCount++ if $subsys=~/Z/;
#print "NEW PREFIX: $newPrefixDate NEW FILE: $newOutputFile\n";
# set playback timeframe for the file we're about to playback, using the date of the file
# if not specified or that from the from if it is. The start time has already been set
# earlier but when not starting at the beginning, we need to back up 1 interval since the
# first one is never reported.
my $tempDate=($fromDate eq '0') ? $recDate : $fromDate;
$fromSecs=getSeconds($tempDate, $fromTime);
$fromSecs-=$interval if $fromTime!=0;
# The ending time is either the same date as the starting one (unless overriden by the user
# for files that cross midnight) and we need to add a fraction to the ending time in case
# fractional timestamps in file. Max time is Jan 19, 2038 but we'll use Jan 1 if needed.
$tempDate=$thruDate if defined($thruDate) && $thruDate ne'0';
$thruSecs=(!defined($thru)) ? 2145934800 : getSeconds($tempDate, $thruTime).'.999';
# this is just to make debugging time frames easier especially if user gets odd results.
if ($debug & 1)
{
my $fromstamp=getDateTime($fromSecs);
my $thrustamp=getDateTime($thruSecs);
print "PlayBack From: $fromstamp Thru: $thrustamp\n";
}
if ($zInFlag)
{
$ZPLAY=Compress::Zlib::gzopen($file, "rb") or logmsg("F", "Couldn't open '$file'");
}
else
{
open PLAY, "<$file" or logmsg("F", "Couldn't open '$file'");
}
if ($extractMode)
{
my $base=basename($file);
$outfile=(-d $extract) ? "$extract$Sep$base" : "$extract-$base";
#print "BASE: $base PREFIX: $prefix OUT: $outfile\n";
error("--extract specifies an output file with the same name as original!") if $outfile eq $file;
logmsg('I', "Extracting to '$outfile'");
# compress the output file, but only if the input one was compressed.
$ZRAW=Compress::Zlib::gzopen($outfile, 'wc') or logmsg("F", "Couldn't create '$outfile'") if $outfile=~/gz$/;
open RAW, ">$outfile" or logmsg("F", "Couldn't create '$outfile'") if $outfile!~/gz$/;
}
# only call this if generating plot data either in file or on terminal AND
# only one time per output file
if ($plotFlag && ($newOutputFile || $options=~/u/))
{
# Before we do anything else, close any files that were opened last pass
# noting 'closeLogs()' also calls setFlags($subsys)
closeLogs($lastSubsys) if $lastSubsys ne '';
# Open all output files here based on what was in merged subsystems.
setFlags($subsysAll);
print "SetFlags: $subsysAll\n" if $debug & 1;
# If playback file has a prefix before its hostname things get more complicated
# as we want to preserve that prefix and at the same time honor -f.
$filespec=$filename;
if ($prefix=~/(.+)-$Host/)
{
my $temp=$1;
$temp=~s/.*$Sep//;
$filespec.=(-d $filespec) ? "$Sep$temp" : "-$temp";
}
# note we're only passing '$file' along in case we need diagnostics and we're also
# resetting '$subsys' to match ALL the subsystems selected for this set of file(s)
my $saveSubsys=$subsys;
$subsys=$subsysAll;
$newfile=newLog($filespec, $recDate, $recTime, $recSecs, $recTZ, $file);
if ($newfile ne '1')
{
# This is the most common failure mode since people rarely use -ou
# and having 2 separate conditions gives us more flexibility in messages
if ($options!~/u/)
{
print " Plotfile '$newfile' already exists and will not be touched\n";
print " '-oc' to create a new one OR '-oa' to append to it\n";
}
else
{
print " Plotfile '$newfile' exists and is newer than $file\n";
print " You must specify '-ocu' to force creation of a new one\n";
}
next;
}
$subsys=$saveSubsys;
$newOutputFile=0;
$lastSubsys=$subsysAll; # used to track which files were actually opened
}
# when processing data for a new prefix/date and printing on a terminal
# we need to print totals from previous file(s) if there were any and
# reset total. However is --statopts s (as opposed to S), we do subtotals
# for each file
if ($filename eq '' && ($newPrefixDate || $statOpts=~/s/))
{
if ($statsFlag && $numProcessed)
{
printBriefCounters('A') if $statOpts=~/a/;
printBriefCounters('T');
}
$elapsedSecs=0;
resetBriefCounters();
}
# Whenever a from time specified AND we're doing a new prefix, we need to start out in
# skip mode. In all other cases we read ALL the records. Since --from with no date
# applies to all files, that will also trigger starting out in 'skip' mode.
$skip=($fromSecs && ($fromDate==0 || $newPrefixFlag)) ? 1 : 0;
undef($fileFrom);
$firstTime=$firstTime2=1; # tracks int1 and int2 first time processing
$fileThru=$newMarkerWritten=$timestampFlag=$timestampCounter[$rawPFlag]=0;
$fullTime=0; # so we don't get uninit first time we do $microInterval calculation
$bytes=1; # so no compression error on non-zipped files
$numProcessed++; # it's not until we get here that we can say this
while (1)
{
# read a line from either zip file or plain ol' one
last if ( $zInFlag && ($bytes=$ZPLAY->gzreadline($line))<1) ||
(!$zInFlag && !($line=));
# we always skip comments, but in extract mode we need to echo them to output file
if ($line=~/^#/)
{
if ($extractMode)
{
$ZRAW->gzwrite($line) if $outfile=~/gz$/;
print RAW $line if $outfile!~/gz$/;
}
next;
}
# Doncha love special cases? Turns out when reading back process data
# from a PRC file which was created from multiple logs, if a process from
# one log comes up with the same pid as that of an earlier log, there's
# no easy way to tell. Now there is!
writeInterFileMarker()
if $filename ne '' && $prcFileCount>1 && !$newMarkerWritten;
$newMarkerWritten=1;
# if new interval, it really indicates the end of the last one but its
# time is that of the new one so process last interval before saving.
# if this isn't a valid interval marker the file somehow got corrupted
# which was seen one time before flush error handling was put in. Don't
# know if that was the problem or not so we'll keep this extra test.
$timestampFlag=0;
if ($line=~/^>>>/)
{
# after we've processed the data for the interval that DOESN'T print, we
# need to clear the flag that indicates all intervals need to be processed
$firstPass=0 if $timestampCounter[$rawPFlag]==1;
# we need to make sure both $lastSeconds and $newSeconds track BOTH the
# raw and rawp files, if both exist.
# we need to know later on if we're processing a timestamp AND how many we've seen
# because if we hit EOF and only 1 seen, we have not processed a single, full interval.
$timestampFlag=1;
if ($line!~/^>>> (\d+\.\d+) <<=$fromSecs;
# since we're in an inner loop we need a flag
if ($thruSecs && $lastSeconds[$rawPFlag]>$thruSecs)
{
$doneFlag=1;
last;
}
# Always echo timestamp in extract mode when we're processing this interval
if ($extractMode && !$skip)
{
$ZRAW->gzwrite($line) if $outfile=~/gz$/;
print RAW $line if $outfile!~/gz$/;
}
$timestampCounter[$rawPFlag]++ if !$skip;
if ($timestampCounter[$rawPFlag]==1)
{
# If a second (or more) file for same host, are their timstamps consecutive?
# Since we could have a raw/rawp file the way to tell a new file is that
# $newSeconds will be defined.
# If NOT consecutive (or first file for a host), init 'last' variables, noting
# we also need to init if there was a disk configuration change.
$consecutiveFlag=(!$newHostFlag && defined($newSeconds[$rawPFlag]) &&
$thisSeconds==$newSeconds[$rawPFlag] && !$diskChangeFlag) ? 1 : 0;
$newSeconds[$rawPFlag]=$thisSeconds;
if (!$consecutiveFlag)
{
# if not doing raw/rawp files, init everything, otherwise just init the type we're doing
initLast() if ($playback{$prefix}->{flags} & 1)==0;
initLast($rawPFlag) if $playback{$prefix}->{flags} & 1;
$lastSecs[$rawPFlag]=$thisSeconds;
}
print "ConsecFlag: $consecutiveFlag\n" if $debug & 1;
next;
}
$newSeconds[$rawPFlag]=$fullTime=$thisSeconds; # we use '$fullTime' for $microInterval re-calculation
# track from/thru times for each file to be used for -oA in terminal mode
if (!$skip && !$rawPFlag)
{
$fileFrom=$newSeconds[$rawPFlag] if !defined($fileFrom);
$fileThru=$newSeconds[$rawPFlag];
}
# Normally we fall through on a timestamp marker so we can process the interval results
# but in extract we don't want to generate any output, just record data.
next if $extractMode;
}
next if $skip;
print $line if $debug & 4;
if ($grepPattern ne '')
{
if ($line=~/$grepPattern/)
{
$firstTime=0; # to indicate something found
my $msec=(split(/\./, $newSeconds[$rawPFlag]))[1];
my ($ss, $mm, $hh, $mday, $mon, $year)=localtime($newSeconds[$rawPFlag]);
$datetime=sprintf("%02d:%02d:%02d", $hh, $mm, $ss);
$datetime=sprintf("%02d/%02d %s", $mon+1, $mday, $datetime) if $options=~/d/;
$datetime=sprintf("%04d%02d%02d %s", $year+1900, $mon+1, $mday, $datetime) if $options=~/D/;
$datetime.=".$msec" if ($options=~/m/);
print "$datetime $line";
}
next;
}
if ($extractMode)
{
# clear flag to prevent error message later.
$firstTime=0;
$ZRAW->gzwrite($line) if $outfile=~/gz$/;
print RAW $line if $outfile!~/gz$/;
next;
}
# Either we're processing a timestamp marker OR data entries
# When using a single raw file that has interval markers for all record and newer rawp
# files that only have them for interval2 only we need to force the 'print' flag each time
$interval2Print=1 if $rawPFlag && $recVersion ge '3.3.5';
if ($timestampFlag)
{
# We already skipped first interval marker. As for the second one, which indicates the end of
# a complete set of data, we only process that if we have consecutive files in which case
# we get to use the last file's data for the previous interval's data. BUT we have to make
# sure 'initInterval' called for second interval which may have been skipped.
my $saveI2P=$interval2Print; # gets reset to 0 during intervalEnd()
intervalEnd($lastSeconds[$rawPFlag]) if $consecutiveFlag || $timestampCounter[$rawPFlag]>2;
initInterval() if $timestampCounter[$rawPFlag]==2;
$firstTime2=0 if $saveI2P;
}
else
{
dataAnalyze($subsys, $line);
}
$firstTime=0;
}
# Write 'next' timestamp at end of file.
if ($extractMode)
{
$ZRAW->gzwrite($line) if $outfile=~/gz$/;
print RAW $line if $outfile!~/gz$/;
}
# We really only need message when -p specifies single file
if ($firstTime && $numSelected==1)
{
print "No records selected for playback! Are --from/--thru` wrong?\n";
next;
}
# normally samples will end on timestamp marker (even if last interval) and therefore
# processed above. However in pre-3.1.2 releases timestamps weren't written when
# logs rolled and so we need to process the last interval in those cases as well.
intervalEnd($newSeconds[$rawPFlag]) if !$timestampFlag && $recVersion lt '3.1.2';
my $tmpsys=$subsys;
$tmpsys=~s/[YZ]//g;
$ZPLAY->gzclose() if $zInFlag;
close PLAY if !$zInFlag;
# if we reported data from this file (we may have skipped it entirely if --from
# used with multiple files), calculate how many seconds reported on in for
# stats reporting with -oA, but only if at least 1 full interval processed
if (!$skip && !$rawPFlag && !$extractMode && $timestampCounter[$rawPFlag]>1)
{
# Note that by default we never include first interval data, but if this was a
# consecitive file we need to include that interval to so add it back in
$playbackSecs=$fileThru-$fileFrom;
$playbackSecs+=$interval if $consecutiveFlag;
$elapsedSecs+=$playbackSecs;
}
# for easier reading...
print "\n" if $debug & 1;
# This should be pretty rare..
logmsg("E", "Error reading '$file'\n") if $bytes==-1;
last if $doneFlag;
}
# Close logs that are open from last pass
closeLogs($lastSubsys) if $numProcessed;
# Always print last set of summary data...
printProcAnalyze() if $procAnalCounter;
printSlabAnalyze() if $slabAnalCounter;
# if printing to terminal, be sure to print averages & totals for last file processed
if (!$rawPFlag && $statsFlag && $filename eq '')
{
$subsys=$subsysAll; # in case mixed raw/rawp we need to reset
$subsys=~s/y//;
printBriefCounters('A') if $statOpts=~/a/;
printBriefCounters('T');
}
`stty echo` if !$PcFlag && $termFlag && !$backFlag; # in brief mode, we turned it off
my $temp=(!$msgFlag) ? ' Try again with -m.' : '';
print "No files selected contain the selected data.$temp\n" if !$numProcessed;
exit(0);
}
###########################
# R e c o r d M o d e
###########################
# Would be nice someday to migrate all record-specific checks here
error("-offsettime only applies to playback mode") if defined($offsetTime);
# need to load even if interval is 0, but don't allow for -p mode
loadPids($procFilt) if $subsys=~/Z/;
loadUids($passwdFile) if $subsys=~/Z/;
# In case running on a cluster, record the name of the host we're running on.
# Track in collecl's log as well as syslog
my $nohup=($nohupFlag && !$daemonFlag) ? '[--nohup]' : ''; # only announce --nohup if not daemon
my $temp=($runas ne '') ? "as user '$runas' " : '';
$temp.=($pname ne '') ? "(running as '$pname') " : '';
$message="V$Version Beginning execution$nohup ${temp}on $myHost...";
logmsg("I", $message);
logsys($message);
checkHiRes() if $daemonFlag; # check for possible HiRes/glibc incompatibility
# now let's report any messages that occurred earlier
foreach my $message (@messages)
{
my ($severity, $text)=split(/-/, $message, 2);
logmsg($severity, $text);
}
# initialize. noting if the user had only selected subsystems not supported
# on this platform, initRecord() will have deselected them!
initRecord();
error("no subsystems selected") if $subsys eq '' && $import eq ''; # ok in --import mode
# Process I/O stats are a little tricky. initRecord() sets $processIOFlag based on kernel's
# capabilities, but if user has disabled them, we then need to clear that flag.
error("process I/O features not enabled in this kernel") if $procOpts=~/i/i && !$processIOFlag;
error("process options i and I are mutually exclusive") if $procOpts=~/i/ && $procOpts=~/I/;
error("you cannot use --top and IO options with this kernel") if $topIOFlag && !$processIOFlag;
error("you cannot use --top and IO options with --procopt I") if $topIOFlag && $procOpts=~/I/;
$processIOFlag=0 if $procOpts=~/I/;
if ($subsys=~/y/i && !$slabinfoFlag && !$slubinfoFlag)
{
logmsg("W", "Slab monitoring disabled because neither /proc/slabinfo nor /sys/slab exists");
$yFlag=$YFlag=0;
$subsys=~s/y//ig;
}
# We can't do this until we know if the data structures exist.
loadSlabs($slabFilt) if $subsys=~/y/i;
# now that subsys accurately reflects the systems we're collectling data on we
# can safely initialize out export if one is defined.
if ($expName ne '')
{
my $initName="${expName}Init";
&$initName(@expOpts);
}
# this sets all the xFlags as specified by -s. At least one must be set to
# write to the 'tab' file.
setFlags($subsys);
# In case displaying output. We also need the recorded version to match ours.
initFormat();
initLast();
$recVersion=$Version;
# This has to go after initFormat() since it loads '$envRules' and may init
# stuff needed by printTerm()
if ($envTestFile ne '')
{
envTest();
exit(0);
}
# Since we have to check subsystem specific options against data in recorded
# file, let's not do it twice, but we have to do it AFTER initFormat()
checkSubsysOpts();
# L a s t M i n u t e V a l i d a t i o n
# This needs to be done after loadConfig and only in record mode
logmsg('W', "Couldn't find 'ipmitool' in '$ipmitoolPath'")
if $subsys=~/E/ && $Ipmitool eq '';
# These can only be done after initRecord()
# Since it IS possible for a server to be running as an MDS and a client, we need the following
error("-sL applies to a server only running as an MDS when used with --lustopts D")
if $subsys=~/L/ && $NumMds && !$CltFlag && $lustOpts!~/D/;
error("--lustopts D only applies to HP-SFS")
if $lustOpts=~/D/ && $sfsVersion eq '';
if ($options=~/x/i)
{
error("exception reporting requires --verbose")
if !$verboseFlag;
error("exception reporting only applies to -sD and lustre OST details or MDS/Client summary")
if ($subsys!~/[DLl]/ || ($subsys=~/L/ && $NumOst==0) ||
($subsys=~/l/ && $NumMds+$CltFlag==0));
error("exception reporting must be to a terminal OR a file in -P format")
if ($filename ne "" && !$plotFlag) || ($filename eq "" && $plotFlag);
}
# L a s t M i n u t e C h a n g e s T o F o r m a t t i n g
# OK, so it's getting messy. The decision to use brief/verbose is made in setOutputFormat()
# but it's called much earlier, certainly before if we know what types of lustre node that
# gets determined in initFormat() which gets called up above. Perhaps over time other
# last minute tests will need a home and this may prove to be it.
# The purpose of this is that in verbose mode when a single type of data is being displayed
# we'll have set $sameColsFlag, but now that we know more about the lusre configuration
# we may have to clear that setting.
if ($subsys=~/l/i && $verboseFlag)
{
$sameColsFlag=0 if $CltFlag+$OstFlag+$MdsFlag>1;
}
# daemonize if necessary
if ($daemonFlag)
{
# We need to make sure no terminal I/O
open STDIN, '/dev/null' or logmsg("F", "Can't read /dev/null: $!");
open STDOUT, '>/dev/null' or logmsg("F", "Can't write to /dev/null: $!");
open STDERR, '>/dev/null' or logmsg("F", "Can't write to /dev/null: $!");
# fork a child and exit parent, but make sure fork really works
defined(my $pid=fork()) or logmsg("F", "Can't fork: $!");
exit(0) if $pid;
# Make REALLY sure we're disassociated
setsid() or logmsg("F", "Couldn't setsid: $!");
open STDIN, '/dev/null' or logmsg("F", "Can't read /dev/null: $!");
open STDOUT, '>/dev/null' or logmsg("F", "Can't write to /dev/null: $!");
open STDERR, '>/dev/null' or logmsg("F", "Can't write to /dev/null: $!");
`echo $$ > $PidFile`;
# Now that we're set up to start, if '--runas' has been sprecified we need to do a
# few things that require privs before actually changing our UID. Also note the
# GID is optional.
if ($runas ne '')
{
# we have to make sure the owner ship of the message log is correct.
# This is only an issue for the msglog when a new file gets created to log the first
# messge of the month and we've restarted as root. Steal the code from logmsg() to
# build its name.
($ss, $mm, $hh, $day, $mon, $year)=localtime(time);
$yymm=sprintf("%d%02d", 1900+$year, $mon+1);
$logname=(-d $filename) ? $filename : dirname($filename);
$logname.="/$myHost-collectl-$yymm.log";
`chown $runasUid $logname`;
`chgrp $runasGid $logname` if defined($runasGid);
# now we can change our process's ownership taking care to do the group first
# since we won't be able to change anything once we change our UID.
$EGID=$runasGid if defined($runasGid);
$EUID=$runasUid;
}
}
######################################################
#
# ===> WARNING: No Writing to STDOUT beyond <=====
# since we may be daemonized!
#
######################################################
$SIG{"INT"}=\&sigInt; # for ^C
$SIG{"TERM"}=\&sigTerm; # default kill command
$SIG{"USR1"}=\&sigUsr1; # for flushing gz I/O buffers
# to catch collectl's socket I/O errors, noting graphite.ph sets its own handler
$SIG{"PIPE"}=\&sigPipe if $address ne '';
$flushTime=($flush ne '') ? time+$flush : 0;
# intervals... note that if no main interval specified, we use
# interval2 (if defined OR if only doing slabs/procs) and if not
# that, interval3. Also, if there is an interval3, interval3 IS defined, so we
# have to compare it to ''. Also note that since newlog() can change subsys
# we need to wait until after we call it to do interval/limit validation.
# be sure to ignore interval error checks for --showcolheader
$origInterval=$interval;
($interval, $interval2, $interval3)=split(/:/, $interval);
if (!$showColFlag)
{
error("interval2 only applies to -s y,Y or Z")
if defined($interval2) && $interval2 ne '' && $subsys!~/[yYZ]/;
error("interval2 must be >= interval1")
if defined($interval) && defined($interval2) && $interval2 ne '' && $interval>$interval2;
}
$interval2=$Interval2 if !defined($interval2);
$interval3=$Interval3 if !defined($interval3);
$interval=$interval2 if $origInterval=~/^:/ || ($subsys=~/^[yz]+$/i && $interval!=0);
$interval=$interval3 if $origInterval=~/^::/;
if ($interval!=0)
{
if ($subsys=~/[yYZ]/)
{
error("interval2 must be >= main interval")
if $interval2<$interval;
error("interval2 must be the same as interval1 in --top mode")
if $numTop && $interval!=$interval2;
$limit2=$interval2/$interval;
error("interval2 must be a multiple of main interval")
if $limit2!=int($interval2/$interval);
}
if ($subsys=~/E/)
{
error("interval3 must be >= main interval")
if $interval3<$interval;
$limit3=$interval3/$interval;
error("interval3 must be a multiple of main interval")
if $limit3!=int($interval3/$interval);
}
}
else
{
# While we don't want any pauses, we also want to limit the number of collections
# to the same number as would be taken during normal activities. The magic here
# is we can only get here is if $userInterval is not null. By default we assume
# the ratios between ints 1/2/3 to be 1/6/30, but if i2 or i3 specified use those
# as the ratios, not actual intervals. eg for -i5:20, use -i0:4
my ($ui, $ui2, $ui3)=split(/:/, $userInterval);
$ui2='' if !defined($ui2);
$ui3='' if !defined($ui3);
$limit2=($ui2 eq '') ? 6 : $ui2;
$limit3=($ui3 eq '') ? 30 : $ui3;
$interval2=$interval3=0;
print "Interval Lim2: $limit2 Lim3: $limit3\n" if $debug & 1;
# make sure no 'bogus network speed' errors
$DefNetSpeed=0;
}
if ($tworawFlag || $groupFlag)
{
error("-G/--group has been replaced with --tworaw") if $groupFlag;
error("--tworaw require BOTH process and non-procss data") if !$recFlag0 || !$recFlag1;
error("--tworaw requires data collection to a file") if $filename eq '';
# DUE to what seems to be a bug in zlib 2.02 (and maybe others), you cannot flush a buffer
# twice in a row w/o writing to it. A shorter interval causes that to happen to rawp.gz.
error("cannot use -F0 with --tworaw when interval1 not equal interval2") if !$flush && $interval!=$interval2;
error("flush time cannot be < process collection interval, when using --tworaw") if $flush && $flush<$interval2;
}
# Note that even if printing in plotting mode to terminal we STILL call newlog
# because that points the LOG, DSK, etc filehandles at STDOUT
# Also, note that in somecase we set non-compressed files to autoflush
$autoFlush=1 if $flush ne '' && $flush<=$interval && !$zFlag;
newLog($filename, "", "", "", "", "") if ($filename ne '' || $plotFlag);
# We want all final runtime parameters defined before doing this
if ($showHeaderFlag && $playback eq '')
{
initRecord();
my $temp=buildCommonHeader(0, undef);
printText($temp);
exit(0);
}
# If HiRes had been loaded and we're NOT doing 'time' tests, we want to
# align each interval via sigalrm. We HAVE to clear doneFlag here rather
# than at loop top because when collectl receives a sigterm it sets the flag
# and we don't want to set it back to 0.
$doneFlag=0;
if ($hiResFlag && $interval!=0)
{
# Default for deamons is to always align to the primary interval
$alignFlag=1 if $daemonFlag;
# sampling is calculated as multiples of a base time and we set that
# time such that our next sample will occur on the next whole second,
# just to make integer sampling align on second boundaries
$AlignInt=$interval;
$BaseTime=(time-$AlignInt+1)*1000000;
# For aligned time we want to align on either the primary interval OR if
# we're monitoring for processes or slabs, on the secondary one. To make
# all sample times align no matter when they were started, we align based
# on a time of 0 which is 00:00:00 on Jan 1, 1970 GMT
if ($alignFlag)
{
$AlignInt=($subsys=~/[yz]/i) ? $interval2 : $interval;
$BaseTime=0;
}
# Point to our alarm handler and set up some u-constants
$SIG{"ALRM"}=\&sigAlrm;
$uInterval=$interval*1000000;
# Now we can enable our alarm and sleep for at least a full interval, from
# which we'll awake by a 'sigalrm'. The first time thought is based on our
# alignment, which may be '$interval2', but after that it's always '$interval'
# Also note use of arg2 to note first call since arg1 always set to 'ALRM'
# when it fires normally.
$uAlignInt=$AlignInt*1000000;
sigAlrm(undef, 1);
sleep $AlignInt+1;
$uAlignInt=$uInterval;
sigAlrm(); # we're now aligned so reset timer
}
if ($debug & 1 && $options=~/x/i)
{
$temp=$limBool ? "AND" : "OR";
print "Exception Processing In Effect -- SVC: $limSVC $temp IOS: $limIOS ".
"LusKBS: $LimLusKBS LusReints: $LimLusReints\n"
}
# remind user we always wait until second sample before producing results
# if only yY, Z or E or both, we don't wait for the standard interval
$temp=$interval;
$temp=$interval2 if $subsys=~/^[EyYZ]+$/;
$temp=$interval3 if $subsys eq 'E';
print "waiting for $temp second sample...\n" if $filename eq "" && !$quietFlag;
# Need to make sure proc's and env's align with printing of other vars first
# time. In other words, do the first read immediately.
$counted2=($subsys=~/[yYZ]/) ? $limit2-1 : 0;
$counted3=($subsys=~/E/) ? $limit3-1 : 0;
# Figure out how many intervals we want to check for lustre config changes,
# noting that in the debugging case where the interval is 0, we calculate it
# based on a day's worth of seconds.
$lustreCheckCounter=0;
$lustreCheckIntervals=($interval!=0) ?
int($lustreConfigInt/$interval) : int($count/(86400/$lustreConfigInt));
$lustreCheckIntervals=1 if $lustreCheckIntervals==0;
print "Lustre Check Intervals: $lustreCheckIntervals\n" if $debug & 8;
# Looks like only HP-SFS should skip leading 7 fields of client OST data
my $lustreCltOstSkip=($sfsVersion ne '') ? 7 : 0;
# Same thing (sort of) for interconnect interval
$interConnectCounter=0;
$interConnectIntervals=($interval!=0) ?
int($InterConnectInt/$interval) : int($count/(86400/$InterConnectInt));
$interConnectIntervals=1 if $interConnectIntervals==0;
print "InterConnect Interval: $interConnectIntervals\n" if $debug & 2;
if ($options=~/i/)
{
my $temp=buildCommonHeader(0, undef);
printText($temp);
}
# Wait until the last minute to set up the scrolling region so if we crap out
# earlier we haven't screwed up the terminal.
printf "%c[3;%dr", 27, $scrollEnd if $scrollEnd;
# M a i n P r o c e s s i n g L o o p
# This is where efficiency really counts
# $lastHour lets us figure out when it's a new day
my $lastHour=(localtime($rollSecs))[2];
my $lastFirstPid=0;
for (; $count!=0 && !$doneFlag; $count--)
{
# When in server mode we always need to check for readable socket
# but be sure to do without any timeout
if ($serverFlag)
{
print "Looking for connection...\n" if $debug & 64;
if ($newHandle=($select->can_read(0))[0])
{
print "Socket 'can read'\n" if $debug & 64;
if ($newHandle==$sockServer)
{
$socket=$sockServer->accept() || logmsg('F', "Couldn't accept socket request. Reason: $!");
$select->add($socket);
push @sockets, $socket;
my $client=inet_ntoa((sockaddr_in(getpeername($socket)))[1]);
logmsg('I', "New socket connection from $client");
}
else
{
$socket=$newHandle;
my $message=<$socket>;
if (!defined($message))
{
logmsg('W', "Client closed socket");
for (my $i=0; $iremove($socket);
$socket->close();
}
else
{
print "Received: $message" if $debug & 64;
}
}
}
}
# Use the same value for seconds for the entire cycle
if ($hiResFlag)
{
# we have to fully qualify name because or 'require' vs 'use'
($intSeconds, $intUsecs)=Time::HiRes::gettimeofday();
}
else
{
$intSeconds=time;
$intUsecs=0;
}
# T i m e F o r a N e w L o g ?
if ($logToFileFlag && $rollSecs)
{
# if time to roll, do so and recalculate next roll time.
if ($intSeconds ge $rollSecs)
{
# if new day, do day-level inits of which there aren't all that many
($sec, $min, $hour, $day, $mon, $year)=localtime($rollSecs);
initDay() if $hour<$lastHour;
$lastHour=$hour;
# We need to make sure each logfile has headers. Since this flag is used interactively
# as well we can't just clear it here.
$zlibErrors=$headersPrinted=0;
newLog($filename, "", "", "", "", "");
$rollSecs+=$rollIncr*60;
# Just like the logic above to calculate the time of our first roll, we
# need to see if we're going to cross a time change boundary
if ($rollIncr>60)
{
#print "EXP: $expectedHour HOUR: $hour\n";
my $diff=($expectedHour-$hour);
$diff=1 if $diff==-23;
$rollSecs+=$diff*3600;
$expectedHour+=$rollIncr/60;
$expectedHour%=24;
logmsg("I", "Time change! Did you remember to change your watch?") if $diff!=0;
}
logmsg("I", "Logs rolled");
}
}
# G a t h e r S T A T S
# This is the section of code that needs to be reasonably efficient. but first, start
# the interval with a time marker noting we have to first make sure we're padding with
# 0's, then truncate to 2 digit precision noting is a rawp file we only write a marker
# during interval2
$counted2++;
$fullTime=sprintf("%d.%06d", $intSeconds, $intUsecs);
record(1, sprintf(">>> %.3f <<<\n", $fullTime)) if $recFlag0;
record(1, sprintf(">>> %.3f <<<\n", $fullTime), undef, 1) if $recFlag1 && $counted2==$limit2;
##############################################################
# S t a n d a r d I n t e r v a l P r o c e s s i n g
##############################################################
if ($bFlag || $BFlag)
{
getProc(0, "/proc/buddyinfo", "buddy");
}
if ($cFlag || $CFlag || $dFlag || $DFlag)
{
# Too crazy to do in getProc() though maybe someday should be moved there
open PROC, ")
{
last if $line=~/^kstat/;
record(2, $line)
if (( ($cFlag || $CFlag) && $line=~/^cpu|^ctx|^proc/) ||
( $DFlag && !$hiResFlag && $line=~/^cpu /));
record(2, "$1\n")
if ($cFlag || $CFlag) && $line=~/(^intr \d+)/;
}
close PROC;
}
if ($jFlag || $JFlag)
{
getProc(0, '/proc/interrupts', 'int', 1);
}
if ($dFlag || $DFlag)
{
getProc(9, "/proc/diskstats", "disk");
}
if ($cFlag || $CFlag)
{
getProc(0, "/proc/loadavg", "load");
}
if ($tFlag || $TFlag)
{
getProc(20, "/proc/net/netstat", 'tcp') if $tcpFilt=~/[IT]/;
getProc(21, "/proc/net/snmp", 'tcp') if $tcpFilt=~/[cimtu]/;
}
if ($iFlag)
{
getProc(0, "/proc/sys/fs/dentry-state", "fs-ds") if $dentryFlag;
getProc(0, "/proc/sys/fs/inode-nr", "fs-is") if $inodeFlag;
getProc(0, "/proc/sys/fs/file-nr", "fs-fnr") if $filenrFlag;
}
if ($lFlag || $LFlag || $lustOpts=~/O/)
{
# Check to see if any services changed and if they did, we may need
# a new logfile as well.
if (++$lustreCheckCounter==$lustreCheckIntervals)
{
newLog($filename, "", "", "", "", "")
if lustreCheckClt()+lustreCheckOst()+lustreCheckMds()>0 && $filename ne '';
$lustreCheckCounter=0;
}
# This data actually applies to both MDS and OSS servers and if
# both services are running on the same node we're only going to
# want to collect it once.
if ($lustOpts=~/D/ && ($NumMds || $OstFlag))
{
my $diskNum=0;
foreach my $diskname (@LusDiskNames)
{
# Note that for scsi, we read the whole thing and for cciss
# quit when we see the line with 'index' in it. Also note that
# for sfs V2.2 we need to skip more for cciss than sd disks
$diskSkip=($sfsVersion lt '2.2' || $LusDiskDir=~/sd_iostats/) ? 2 : 14;
$statfile="$LusDiskDir/$diskname";
getProc(2, $statfile, "LUS-d_$diskNum", $diskSkip, undef, 'index');
$diskNum++;
}
}
# OST Processing
if ($OstFlag)
{
# Note we ALWAYS read the base ost data
for ($ostNum=0; $ostNum<$NumOst; $ostNum++)
{
$dirspec="/proc/fs/lustre/obdfilter/$lustreOstSubdirs[$ostNum]";
getProc(1, "$dirspec/stats", "OST_$ostNum", undef, undef, "^io");
# for versions of SFS prior to 2.2, there are only 9 buckets of BRW data.
getProc(2, "$dirspec/brw_stats", "OST-b_$ostNum", 4, $numBrwBuckets)
if $lustOpts=~/B/;
}
}
# MDS Processing
if ($NumMds)
{
my $type=($cfsVersion lt '1.6.0') ? 'MDT' : 'MDS';
getProc(3, "/proc/fs/lustre/mdt/$type/mds/stats", "MDS");
}
# CLIENT Processing
if ($CltFlag)
{
$fsNum=0;
foreach $subdir (@lustreCltDirs)
{
# For vanilla -sl we only need read/write info, but lets grab metadata file
# we're at it. In the case of --lustopts R, we also want readahead stats
getProc(11, "/proc/fs/lustre/llite/$subdir/stats", "LLITE:$fsNum", 1, 19);
getProc(0, "/proc/fs/lustre/llite/$subdir/read_ahead_stats", "LLITE_RA:$fsNum", 1)
if $lustOpts=~/R/;
$fsNum++;
}
# RPC stats are optional for both clients and servers
if ($lustOpts=~/B/)
{
for ($index=0; $index<$NumLustreCltOsts; $index++)
{
getProc(2, "$lustreCltOstDirs[$index]/rpc_stats", "LLITE_RPC:$index", 8, 11);
}
}
# Client OST detail data
if ($lustOpts=~/O/)
{
for ($index=0; $index<$NumLustreCltOsts; $index++)
{
getProc(12, "$lustreCltOstDirs[$index]/stats", "LLDET:$index ", $lustreCltOstSkip);
}
}
}
}
# even if /proc not there (nothing exported/mounted), it could
# show up later so we need to be sure and look every time
if ($fFlag || $FFlag)
{
getProc(8, '/proc/net/rpc/nfs', "nfsc-") if $nfsCFlag;
getProc(8, '/proc/net/rpc/nfsd', "nfss-") if $nfsSFlag;
}
if ($mFlag)
{
getProc(0, "/proc/meminfo", "");
getProc(5, "/proc/vmstat", "");
}
# NOTE - unlike other detail data this is only recorded when explicitly requested
if ($MFlag)
{
for (my $i=0; $i<$CpuNodes; $i++)
{
# skip first line which is blank, noting 'Node X' is already part of data
getProc(0, "/sys/devices/system/node/node$i/meminfo", "numai", 1);
# only if we want hits, adds about 2 secs for 2 node home box
getProc(0, "/sys/devices/system/node/node$i/numastat", "numas Node $i");
}
}
if ($sFlag)
{
getProc(0, "/proc/net/sockstat", "sock");
}
if ($nFlag || $NFlag)
{
if ($rawNetFilter eq '' && $rawNetIgnore eq '')
{ getProc(0, "/proc/net/dev", "Net", 2); }
else
{ getProc(7, "/proc/net/dev", "Net", 2); }
}
if ($xFlag || $XFlag)
{
# Whenever we hit the end of interconnect checking interval we need to
# see if any of them changed configuration (such as an IB port fail-over)
# NOTE - we do the $filename test last so we ALWAYS do the elan/ib checks
# even if printing to terminal.
if (++$interConnectCounter==$interConnectIntervals)
{
newLog($filename, "", "", "", "", "")
if (($quadricsFlag && elanCheck()) || ($mellanoxFlag && ibCheck()))
&& $filename ne '';
$interConnectCounter=0;
}
# only if there is indeed quadric stats detected
if ($quadricsFlag && $NumXRails)
{
for ($i=0; $i<$NumXRails; $i++)
{
getProc(0, "/proc/qsnet/ep/rail$i/stats", "Elan$i");
}
}
if ($mellanoxFlag && $NumHCAs)
{
for ($i=0; $i<$NumHCAs; $i++)
{
if ( -e $SysIB )
{
if ( -e $PQuery )
{
foreach $j (1..2)
{
if ($HCAPorts[$i][$j]) # Make sure it has an active port
{
getExec(1, "$PQuery -r $HCALids[$i][$j] $j 0xf000", "ib$i-$j");
}
}
}
}
elsif ( -e $VoltaireStats )
{
# If Voltaire ever supports multiple HCAs, we'll need the
# uncommented code instead
getProc(0, $VoltaireStats, 'ib0', 3, 2);
#getProc(0, "/proc/voltaire/ib$i/stats", "ib$i", 3, 2);
}
else
{
# Currently only 1 port is active, but if more are, we need to
# deal with them
foreach $j (1..2)
{
if ($HCAPorts[$i][$j]) # Make sure it has an active port
{
# Grab counters and do an immediate reset of them
getExec(2, "$PCounter -h $HCAName[$i] -p $j", "ib$i-$j");
`$PCounter -h $HCAName[$i] -p $j -s 5 >/dev/null`;
}
}
}
}
}
}
# Custom data import
logdiag("begin import data") if ($utimeMask & 1) && $impNumMods;
for (my $i=0; $i<$impNumMods; $i++) { &{$impGetData[$i]}(); }
logdiag("interval1 done") if $utimeMask & 1;
#############################################
# I n t e r v a l 2 P r o c e s s i n g
#############################################
if (($yFlag || $YFlag || $ZFlag) && $counted2==$limit2)
{
if ($yFlag || $YFlag)
{
# NOTE - $SlabGetProc is either 99 for all slabs or 14 for selective
if ($slabinfoFlag)
{
getProc($SlabGetProc, "/proc/slabinfo", "Slab", 2);
}
else
{
# Reading the whole directory and skipping links via the 'skip' hash
# is only about about 1/2 second slower over the day so let's just do it.
opendir SLUBDIR, "/sys/slab" or logmsg('E', "Couldn't open '/sys/slub'");
while ($slab=readdir SLUBDIR)
{
next if $slab=~/^\./;
next if $slabFilt ne '' && !defined($slabdata{$slab});
next if defined($slabskip{$slab});
# See if a new slab appeared, noting this doesn't apply when using
# --slabfilt because of the optimization 'next' for '$slabFilt' above
# also remember since we're only looking at root slabs, we'll never
# discover 'linked' ones
if (!defined($slabdata{$slab}))
{
$newSlabFlag=1;
logmsg("W", "New slab detected: $slab");
}
# Whenever there are 'new' slabs to read (which certainly includes the first
# full pass or any time we change log files) read constants before reading
# variant data.
getSys('Slab', '/sys/slab', $slab, 1, ['object_size', 'slab_size', 'order','objs_per_slab'])
if $firstPass || $newRawSlabFlag || $newSlabFlag;
getSys('Slab', '/sys/slab', $slab, 1, ['objects', 'slabs']);
$newSlabFlag=0;
}
}
}
if ($ZFlag)
{
# need to know when we're looking at the first proc of this interval
$firstProcCycle=1;
# Process Monitoring RULES
# if --procopt p OR --procfilt p and only pids
# - only look at pids in %pidProc and nothing more
# - if + and no --procopt t, never look for new threads
# - if --procopt t, always look for new threads whether + or not
# else always look for new processes
# - if --procopt p look for threads for each pid
undef %pidSeen;
if ($pidOnlyFlag)
{
foreach $pid (keys %pidProc)
{
# When looking at threads, we read ALL data from /proc/pid/task/pid
# rather than /proc/pid so we can be assured we only seeing stats
# for the main process. Later on too...
# But also note earliest kernels only support process io under /proc/pid
$task=$taskio=($allThreadFlag || $oneThreadFlag) ? "$pid/task/" : '';
$taskio='' if !-e "/proc/$pid/task/$pid/io";
# note that not everyone has 'Vm' fields in status so we need
# special checks. Also note both here and below whenever we process a pid
# and not --procopt p (we could have gotten here via --procfilt p...) and
# we're doing threads on this pid, see if any new threads showed up. If
# this gets much more involved it should probably become a sub since we do
# it below too.
$pidSeen{$pid}=getProc(17, "/proc/$task/$pid/stat", "proc:$pid stat", undef, 1);
$pidSeen{$pid}=getProc(13, "/proc/$task/$pid/status", "proc:$pid")
if $pidSeen{$pid}==1;
$pidSeen{$pid}=getProc(16, "/proc/$task/$pid/cmdline", "proc:$pid cmd", undef, 1)
if $pidSeen{$pid}==1;
$pidSeen{$pid}=getProc(17, "/proc/$taskio/$pid/io", "proc:$pid io")
if $pidSeen{$pid}==1 && $processIOFlag && ($rootFlag || -r "/proc/$taskio/$pid/io");
findThreads($pid) if $allThreadFlag || ($oneThreadFlag && $procOpts!~/p/ && $pidThreads{$pid});
}
}
else
{
opendir DIR, "/proc" or logmsg("F", "Couldn't open /proc");
while ($pid=readdir(DIR))
{
next if $pid=~/^\./; # skip . and ..
next if $pid!~/^\d/; # skip not pids
next if defined($pidSkip{$pid});
next if !defined($pidProc{$pid}) && pidNew($pid)==0;
# see comment in previous block
$task=$taskio=($allThreadFlag || $oneThreadFlag) ? "$pid/task/" : '';
$taskio='' if !-e "/proc/$pid/task/$pid/io";
print "%%% READPID $pid\n" if $debug & 256;
$pidSeen{$pid}=getProc(17, "/proc/$task/$pid/stat", "proc:$pid stat", undef, 1);
$pidSeen{$pid}=getProc(13, "/proc/$task/$pid/status", "proc:$pid")
if $pidSeen{$pid}==1;
$pidSeen{$pid}=getProc(16, "/proc/$task/$pid/cmdline", "proc:$pid cmd", undef, 1)
if $pidSeen{$pid}==1;
$pidSeen{$pid}=getProc(17, "/proc/$taskio/$pid/io", "proc:$pid io")
if $pidSeen{$pid}==1 && $processIOFlag && ($rootFlag || -r "/proc/$taskio/$pid/io");
findThreads($pid) if $allThreadFlag || ($oneThreadFlag && $procOpts!~/p/ && $pidThreads{$pid});
}
}
# if --procopts t OR '+' with --procfilt
if ($allThreadFlag || $oneThreadFlag)
{
foreach $pid (keys %tpidProc)
{
# Location of thread stats is below parent, but I/O only there when kernel patched!
$task=$taskio=($allThreadFlag || $oneThreadFlag) ? "$pid/task/" : '';
$taskio='' if !-e "/proc/$pid/task/$pid/io";
# The 'T' lets the processing code know it's a thread for formatting purposes
$tpidSeen{$pid}=getProc(17, "/proc/$task/$pid/stat", "procT:$pid stat", undef, 1);
$tpidSeen{$pid}=getProc(13, "/proc/$task/$pid/status", "procT:$pid")
if $tpidSeen{$pid}==1;
$tpidSeen{$pid}=getProc(17, "/proc/$taskio/$pid/io", "procT:$pid io")
if $tpidSeen{$pid}==1 && $processIOFlag && ($rootFlag || -r "/proc/$taskio/$pid/io");
}
}
# how else will we know if a process exited?
# This will also clean up stale thread pids as well.
cleanStalePids();
}
$counted2=0;
logdiag("interval2 done") if $utimeMask & 1;
}
#############################################
# I n t e r v a l 3 P r o c e s s i n g
#############################################
if ($EFlag && ++$counted3==$limit3)
{
# On the off chance someone deleted it (how do you say overkill?)
if (!-e $IpmiCache)
{
logmsg('E', "Who deleted my cache file '$IpmiCache'?");
logmsg('I', "Recreated missing cache file");
$command="$Ipmitool sdr dump $IpmiCache";
`$command`;
}
# About the same overhead to invoke ipmitool twice but much less elapsed time.
getExec(3, "$Ipmitool -c -S $IpmiCache exec $ipmiExec", 'ipmi');
$counted3=0;
logdiag("interval3 done") if $utimeMask & 1;
}
###########################################################
# E n d O f I n t e r v a l P r o c e s s i n g
###########################################################
# if printing to terminal OR generating data in plot format (or both)
# we need to wait until the end of the interval so complete data is in hand
if (!$logToFileFlag || $plotFlag || $export ne '')
{
$fullTime=sprintf("%d.%06d", $intSeconds, $intUsecs);
intervalEnd(sprintf("%.3f", $fullTime));
logdiag('interval processed') if $utimeMask & 1;
}
# If there was a disk configuration change and writing to plot files (changes
# can't be detected when writing to raw file), create new log files.
if ($diskChangeFlag && $plotFlag && $filename ne '')
{
logmsg('I', 'Creating new log file') if $options=~/u/;
logmsg('W', 'all data mixed in same file! use -ou to force unique files!') if $options!~/u/;
newLog($filename, "", "", "", "", "");
}
$diskChangeFlag=0;
# If our parent's pid went away we're done, unless --nohup specified or we're a daemon
if (!-e "/proc/$myPpid" && !$daemonFlag && !$nohupFlag)
{
logmsg('W', 'parent exited and --nohup not specified');
last;
}
# if we'll pass the end time while asleep, just get out now.
last if $endSecs && ($intSeconds+$interval)>$endSecs;
# NOTE - I tried used select() as timer when no HiRes but got premature
# wakeups on early 2.6 testing and so went back to sleep(). Also, in
# case we lose our wakeup signal, only sleep as long as requested noting
# we SHOULD get woken up before this timer expires since we already used
# up part of our interval with data collection
flushBuffers() if !$autoFlush && $flushTime && time>=$flushTime;
if ($interval!=0)
{
sleep $interval if !$hiResFlag;
Time::HiRes::usleep($uInterval) if $hiResFlag;
}
$firstPass=0;
$newRawSlabFlag=0 if $counted2==0; # interval 2 just processed
next;
}
# the only easy way to tell a complete interval is by writing a marker, with
# not time, since we don't need it anyway.
if ($hiResFlag)
{
($intSeconds, $intUsecs)=Time::HiRes::gettimeofday();
$fullTime=sprintf("%d.%06d", $intSeconds, $intUsecs);
}
else
{
$fullTime=time;
}
record(1, sprintf(">>> %.3f <<<\n", $fullTime)) if $recFlag0;
record(1, sprintf(">>> %.3f <<<\n", $fullTime), undef, 1) if $recFlag1;
# close logs cleanly and turn echo back on because when 'brief' we turned it off.
closeLogs($subsys);
unlink $PidFile if $daemonFlag;
`stty echo` if !$PcFlag && $termFlag && !$backFlag;
printf("%c[r", 27) if $numTop && $userSubsys ne '';
printf "%c[%d;H\n", 27, $scrollEnd+$numTop+2 if $numTop;
logmsg("I", "Terminating...");
logsys("Terminating...");
sub preprocSwitches
{
my $switches='';
foreach $switch (@ARGV)
{
# Cleaner to not allow -top and force --top
error("invalid switch '$switch'. did you mean -$switch?")
if $switch=~/^-to/;
# multichar switches COULD be single char switch and option
if ($switch=~/^-/ && length($switch)>2)
{
$use=substr($switch, 0, 2).' '.substr($switch,2);
error("invalid switch '$switch'. did you mean -$switch? if not use '$use'")
if $switch=~/^-al|^-ad|^-be|^-co|-^de|^-de|^-en|^-fl|^-he|^-no|^-in|^-ra/;
error("invalid switch '$switch'. did you mean -$switch? if not use '$use'")
if $switch=~/^-li|^-lu|^-me|-^ni|^-op|^-su|^-ro|^-ru|^-ti|^-wi|^-pr/;
}
$switches.="$switch ";
}
return($switches);
}
# This only effects multiple files for the same system on the same day.
# In most cases, those log files will have been run with the same parameters
# and as a result when their output is simply merged into single 'tab' or
# detail files. However on rare occasions, the configurations will NOT be the
# same and the purpose of this function is to recognize that and change the
# processing parameters according, if possible. The best example of this is
# if one generates one log based on -scd and a second on -scm. By forcing
# the processing of both to be -scdm, the resultant 'tab' file will contain
# everything. Alas, things get more complicated with detail files and even
# more so with lustre detail files if filesystems are mounted/umounted, etc.
# In any event, the details are in the code...
#
# NOTE - if any files cannot be processed, none will and the user will be
# require to change command options
sub preprocessPlayback
{
my $playbackref=shift;
my ($selected, $header, $i);
my ($lastPrefix, $thisSubSys, $thisInterval, $lastInterval, $mergedInterval);
my ($lastSubSys, $lastSubOpt, $lastNfs, $lastDisks, $lastLustreConfig, $lastLustreSubSys);
local ($configChange, $filePrefix, $file);
$selected=0;
$configChange=0;
$lastPrefix=$lastLustreConfig=$lastInterval=$mergedInterval="";
foreach $file (@$playbackref)
{
print "Preprocessing: $file\n" if $debug & 2048;
# need to do individual file checks in case filespec matches bad files
if ($file!~/(.*-\d{8})-\d{6}\.raw[p]*/)
{
$preprocErrors{$file}="I:its name is wrong format";
next;
}
$filePrefix=$1;
$playback{$filePrefix}->{flags}|=0 if !defined($playback{$filePrefix}->{flags});
if (-z $file)
{
$preprocErrors{$file}="I:its size is zero";
next;
}
if ($file!~/raw$|rawp$|gz$/)
{
$preprocErrors{$file}="I:it doesn't end in 'raw', 'rawp' or 'gz'";
next;
}
# If any files in 'gz' format, make sure we can cope.
$zInFlag=0;
if ($file=~/gz$/ && !$zlibFlag)
{
$zInFlag=1;
$preprocErrors{$file}="E:Zlib not installed";
next;
}
# Read header - cleanup code in newlog: see call to getHeader in newLog()
# Set flags based on whether raw or rawp
$header=getHeader($file);
$header=~/SubSys:\s+(\S+)/;
$thisSubSys=$1;
$playback{$filePrefix}->{flags}|=1 if $file=~/\.rawp/;
$playback{$filePrefix}->{flags}|=2 if $file!~/\.rawp/;
# We finally dropped SubOpts and nfsOpts from the header in V3.2.1-5, but not LustOpts
my $subOpts=($header=~/SubOpts:\s+(\S*)\s+Options:/) ? $1 : '';
my $thisNfsOpts= ($header=~/NfsOpts: (\S*)\s*Interval/) ? $1 : $subOpts;
my $thisDisks= ($header=~/DiskNames: (.*)/) ? $1 : '';
my $thisLustOpts=($header=~/LustOpts: (\S*)\s*Services/) ? $1 : $subOpts;
$thisNfsOpts=~s/[BDMORcom]//g; # in case it came from SubOpts remove lustre stuff, 'com' for pre-lustsvc
$thisLustOpts=~s/[234C]//g; # ditto for nfs stuff
$header=~/Interval:\s+(\S+)/;
$thisInterval=$1;
# If user specified '--procopts i' and file doesn't have data, we can't process it
$flags=($header=~/Flags:\s+(\S+)/) ? $1 : '';
if ($procOpts=~/i/ && $flags!~/i/)
{
$preprocErrors{$file}="E:--procopts i requested but data not present in file";
next;
}
# we need to merge intervals if user has selected her own AND set a flag so
# changeConfig() will update %playbackSettings{} correctly
# NOTE - this has never been allowed as -i not allowed in playback
if ($userInterval ne '')
{
$configChange=4; # will cause config change processing AND -m notice
$mergedInterval=mergeIntervals($thisInterval, $mergedInterval);
# on subsequent files, we need to check for interval consistency
if ($filePrefix eq $lastPrefix)
{
print "Merged Intervals: $mergedInterval\n" if $debug & 2048;
my ($int1, $int2, $int3)= split(/:/, $mergedInterval);
my ($uint1, $uint2, $uint3)=split(/:/, $userInterval);
$preprocErrors{$file}="E:common interval '$mergedInterval' not self-consistent"
if (defined($int2) && ($int1>$int2 || int($int2/$int1)*$int1!=$int2)) ||
(defined($int3) && ($int1>$int3 || int($int3/$int1)*$int1!=$int3));
$preprocErrors{$file}="E:common interval '$mergedInterval' has value(s) > $userInterval"
if $uint1<$int1 ||
(defined($uint2) && defined($int2) && $unint2<$int2) ||
(defined($uint3) && defined($int3) && $unint3<$int3);
$preprocErrors{$file}="E:common interval '$mergedInterval' not consistent with $userInterval"
if (int($uint1/$int1)*$int1!=$uint1) ||
(defined($unint2) && defined($int2) && (int($uint2/$int2)*$int2!=$uint2)) ||
(defined($unint3) && defined($int3) && (int($uint3/$int3)*$int3!=$uint3));
}
}
print "File: $file FileSubSys: $thisSubSys NfsOpts: $thisNfsOpts LustOpts: $thisLustOpts\n"
if $debug & 2048;
# note that -s and --lustsvc override anything in the files AND in case -s contained +/-
# we need to do a merge rather than a wholesale replace
$thisSubSys=mergeSubsys($thisSubSys);
$lastLustreConfig=$lustreSvcs if $lustreSvcs ne '';
$lastLustreConfig.='|||';
# it's only if the prefix for this file is the same as the last that
# we have to do all our interval merging and consistency checks.
$selected++;
if ($filePrefix ne $lastPrefix)
{
configChange($lastPrefix, $lastSubSys, $lastLustreConfig, $mergedInterval)
if $lastPrefix ne '';
# New prefix, so initialize for subsequent tests
$newPrefix=1;
$configChange=0;
$mergedInterval='';
# this returns client/server and version or null string
$thisNfs=checkNfs("", $thisSubSys, $thisNfsOpts);
($thisLustreConfig, $lastLustOpts)=checkLustre('', $header, '', $thisLustOpts);
$lastDisks=checkDisks('', $thisDisks) if $thisSubSys=~/D/;
# useful for telling what may have changed
$playback{$filePrefix}->{subsysFirst}=$thisSubSys;
}
else # subsequent files (if any) for same prefix-date
{
# subsystem checks
$newPrefix=0;
$thisSubSys=checkSubSys($lastSubSys, $thisSubSys);
$thisNfs= checkNfs($thisNfs, $thisSubSys, $thisNfsOpts);
($thisLustreConfig, $lastLustOpts)=
checkLustre($lastLustreConfig, $header, $lastLustOpts, $thisLustOpts);
$lastDisks=checkDisks($lastDisks, $thisDisks) if $thisSubSys=~/D/;
}
$lastPrefix=$filePrefix;
$lastSubSys=$thisSubSys;
$lastLustreConfig=$thisLustreConfig;
$playback{$filePrefix}->{subsys}=$thisSubSys;
}
# If multiple files for this prefix processed there are outstanding
# potential changes we need to check for.
configChange($lastPrefix, $lastSubSys, $lastLustreConfig, $mergedInterval)
if $selected && !$newPrefix;
}
# if no -s, return default subsys
# if -s but no +/- return -s
# otherwise merge...
sub mergeSubsys
{
my $default=shift;
my $newSubsys=$default;
if ($userSubsys ne '')
{
if ($userSubsys!~/[+-]/)
{
$newSubsys=$userSubsys;
}
else
{
if ($userSubsys=~/-(.*)/)
{
my $pat=$1;
$pat=~s/\+.*//; # if followed by '+' string
$default=~s/[$pat]//g; # remove matches
}
if ($userSubsys=~/\+(.*)/)
{
my $pat=$1;
$pat=~s/-.*//; # remove anything after '-' string
$default.=$pat; # add matches
}
$newSubsys=$default;
}
}
return($newSubsys);
}
# This purpose of this routine is to look at the intervals from multiple headers
# and figured out what common intervals would be needed to process them all if the
# user wanted to override them. In effect determine the 'least commmon interval',
# only I'm not going to be too precise since virtually all the time these files
# WILL have the same intervals and calculating the LCI will be a lot of work.
sub mergeIntervals
{
my $interval=shift;
my $merged= shift;
my ($mgr1, $mrg2, $mrg3)=split(/:/, $merged);
my ($int1, $int2, $int3)=split(/:/, $interval);
# if any intervals aren't in the merged list, simply move them in
# which will always be the case the first time through
$mrg1=$int1 if !defined($mrg1) || $mrg1 eq '';
$mrg2=$int2 if !defined($mrg2) || $mrg2 eq '';
$mrg3=$int3 if !defined($mrg3) || $mrg3 eq '';
# get least common intervals, but only if new value defined
$mrg1=lci($int1, $mrg1);
$mrg2=lci($int2, $mrg2) if defined($int2);
$mrg3=lci($int3, $mrg3) if defined($int3);
# return the list of merged intervals
$merged=$mrg1;
$merged.=":$mrg2" if defined($mrg2);
$merged.=(defined($mrg2)) ? ":$mrg3" : "::$mrg3" if defined($mrg3);
return($merged);
}
sub lci
{
my $new=shift;
my $old=shift;
$lci=$old;
if ($new>$old)
{
# if a common multiple, use new interval for lci; other return their product
# which will be common but may NOT be the LEAST common!
$lci=($new==int($new/$old)*$old) ? $new : $old*$new;
}
else
{
# same thing only see if $old a multiple of $new
$lci=($old==int($old/$new)*$new) ? $old : $old*$new;
}
}
sub configChange
{
my $prefix= shift;
my $subsys= shift;
my $config= shift;
my $interval=shift;
my ($services, $mdss, $osts, $clts);
my ($i, $type, $names, $temp, $index);
($services, $mdss, $osts, $clts)=split(/\|/, $config);
print "configChange() -- Pre: $prefix Svcs: $services Mds: $mdss Osts: $osts Clts: $clts Int: $interval\n"
if $debug & 8;
# Usually there are no existing messages, but we gotta check...
$index=defined($preprocMessages{$prefix}) ? $preprocMessages{$prefix} : 0;
if ($configChange)
{
$preprocMessages{$prefix.'|'.$index++}=" -s overridden to '$subsys'"
if $configChange & 1;
$preprocMessages{$prefix.'|'.$index++}=" --lustsvr overridden to '$services'"
if $configChange & 2;
$preprocMessages{$prefix.'|'.$index++}=" -i overridden from '$interval' to '$userInterval'"
if $configChange & 4;
foreach $i (8,16,32)
{
next if !($configChange & $i);
if ($i==8) { $types=$mdss; $temp="MDS"; }
if ($i==16) { $types=$osts; $temp="OST"; }
if ($i==32) { $types=$clts; $temp="Client"; }
$preprocMessages{$prefix.'|'.$index++}=" combined Lustre $temp objects now '$types'";
}
$preprocMessages{$prefix}=$index;
$playbackSettings{$prefix}="$subsys|$services|$mdss|$osts|$clts|$interval";
print "Playback -- Prefix: $prefix Settings: $playbackSettings{$prefix}\n" if $debug & 2048;
}
# Send these to log if we're not running interactively and -m not specified
for ($i=0; !$termFlag && !$msgFlag && $i<$index; $i++)
{
logmsg("W", $preprocMessages{$prefix.'|'.$i});
}
return;
}
sub checkSubSys
{
my $lastSubSys=shift;
my $thisSubSys=shift;
my ($nextSubSys, $i);
print "Check SubSys -- Last: $lastSubSys This: $thisSubSys\n"
if $debug & 2048;
# if any differences between 'this' and 'last', we have a config change.
my $temp1=$thisSubSys;
$temp1=~s/[$lastSubSys]//g; # remove 'last' from 'this'
my $temp2=$lastSubSys;
$temp2=~s/[$thisSubSys]//g; # remove 'this' from 'last'
$configChange|=1 if $temp1 ne '' || $temp2 ne '';
# $temp1 contains any NEW subsys in current file, so add them to 'last'
$lastSubSys.=$temp1;
$preprocErrors{$file}="E:-P and details to terminal not allowed"
if $lastSubSys=~/[A-Z]/ && $filename eq '' && $plotFlag;
return($lastSubSys); # has new sub-systems appended
}
sub checkSubsysOpts
{
error("you cannot mix --slabopts with --top") if $slabOpts ne '' && $topSlabFlag;
error("invalid slab option in '$slabOpts'") if $slabOpts ne '' && $slabOpts!~/^[sS]+$/;
error("invalid env option in '$envOpts'") if $envOpts ne '' && $envOpts!~/^[fptCFMT\d]+$/;
$procUsrWidth=8;
if ($procOpts ne '')
{
$procUsrWidth=($procOpts=~s/u(\d+)/u/) ? $1 : 12 if $procOpts=~/u/;
$procCmdWidth=($procOpts=~s/w(\d+)/w/) ? $1 : 1000;
error("minumum username width is 8") if $procUsrWidth<8;
error("invalid process option '$procOpts'") if $procOpts!~/^[cfiIkmprRsStuwxz]+$/;
error("process options i and m are mutually exclusive") if $procOpts=~/i/ && $procOpts=~/m/;
error("your kernel doesn't support process extended info") if $procOpts=~/x/ && !$processCtxFlag;
error("--procopts z can only be used with --top") if !$numTop && $procOpts=~/z/;
}
error("--procstate not one or more of 'DRSTWZ'") if $procState ne '' && $procState!~/^[DRSTWZ]+$/;
# it's possible this is not recognized as running a particular type of service
# from the 'flag's if that service is isn't yet started and so we need
# to check $lustreSvcs too. Be sure to include '$userSubsys' in case -sl was
# specified by the user and then disabled by collectl.
error("--lustsvcs only applies to lustre")
if $lustreSvcs ne '' && $subsys!~/l/i && $userSubsys!~/l/i;
my $cltFlag=($CltFlag || $lustreSvcs=~/c/i) ? 1 : 0;
my $mdsFlag=($MdsFlag || $lustreSvcs=~/m/i) ? 1 : 0;
my $ostFlag=($OstFlag || $lustreSvcs=~/o/i) ? 1 : 0;
error("--lustopts only applies to lustre") if $lustOpts ne '' && $subsys!~/l/i;
error("--lustopts B only applies to Lustre Clts/Osts") if $lustOpts=~/B/ && !$ostFlag && !$cltFlag;
error("--lustopts D only applies to Lustre OSTs/MDSs") if $lustOpts=~/D/ && !$ostFlag && !$mdsFlag;
error("--lustopts M only applies to Lustre Clients") if $lustOpts=~/M/ && !$cltFlag;
error("--lustopts R only applies to Lustre Clients") if $lustOpts=~/R/ && !$cltFlag;
error("--lustopts O only applies to client detail data") if $lustOpts=~/O/ && (!$cltFlag || $subsys!~/L/);
error("you cannot mix --lustopts 'O' with 'M' or 'R'") if $lustOpts=~/O/ && $lustOpts=~/[MR]/;
error("you cannot mix --lustopts 'B' with 'M'") if $lustOpts=~/B/ && $lustOpts=~/M/;
error("you cannot mix --lustopts 'B' with 'R'") if $lustOpts=~/B/ && $lustOpts=~/R/;
# Force if not already specified, but ONLY for details
$lustOpts='BO' if $cltFlag && $subsys=~/L/ && $lustOpts=~/B/ && $lustOpts!~/O/;
}
sub checkNfs
{
my $lastNfs=shift;
my $subsys= shift;
my $subopt= shift;
my $temp;
print "checkNfs(): LastNfs: $lastNfs SubSys: $subsys SubOpt: $subopt\n" if $debug & 2048;
$temp='';
if ($subsys=~/f/i)
{
$temp= ($subopt=~/C/) ? 'C' : 'S';
$temp.=($subopt=~/2/) ? '2' : '3';
}
# all these are legal
return($temp) if $lastNfs eq '';
return($lastNfs) if $temp eq '';
return($lastNfs) if $lastNfs eq $temp; # neither null, both MUST match
# too tricky to handle all possible inconsistencies with multiple files
# so we're only going to print a stock message
$preprocErrors{$filePrefix}="E:confilicting nfs settings with other files of same prefix";
return($temp);
}
sub checkDisks
{
my $lastDisks=shift;
my $thisDisks=shift;
print "checkDisks(): Last: $lastDisks This: $thisDisks\n" if $debug & 2048;
if (($lastDisks ne '') && ($thisDisks ne $lastDisks) && $options!~/u/)
{
$preprocErrors{$filePrefix}="E:confilicting disk names with other files of same prefix and -sD w/o -ou";
}
return($thisDisks);
}
sub checkLustre
{
my $lastConfig= shift;
my $header= shift;
my $lastLustOpts=shift;
my $thisLustOpts=shift;
my ($temp, $thisConfig, $thisMdss, $thisOsts, $thisClts);
my ($services, $mdss, $osts, $clts);
print "checkLustre() -- LastConfig: $lastConfig LastOpts: $lastLustOpts ThisOpts: $thisLustOpts\n"
if $debug & 2048;
($services, $mdss, $osts, $clts)=split(/\|/, $lastConfig);
$services=$osts=$mdss=$clts='' if $lastConfig eq ''; # first time through
# C h e c k L u s t r e S e r v i c e s
# Remember, if set --lustsvcs trumps everything!
if ($lustreSvcs eq '')
{
$thisConfig='';
if ($header=~/MdsNames:\s+(.*)\s*NumOst:\s+\d+\s+OstNames:\s+([^\n\r]*)$/m)
{
# for the first file of a new prefix, we just use the current mdss/osts
# and only check for changes on subsequent calls
if ($1 ne '')
{
$thisMdss=$1;
$thisConfig.='m';
$mdss=($lastConfig eq '') ? $thisMdss : setNames(4, $thisMdss, $mdss);
}
if ($2 ne '')
{
$thisOsts=$2;
$thisConfig.='o';
$osts=($lastConfig eq '') ? $thisOsts : setNames(8, $thisOsts, $osts);
}
}
if ($header=~/CltInfo:\s+(.*)$/m)
{
$thisClts=$1;
$thisConfig.='c';
$clts=($lastConfig eq '') ? $thisClts : setNames(16, $thisClts, $clts);
}
# see if anything new in config
for ($i=0; $i>>>>>>>>>>>> Preproc Error: $errorText\n" if $debug & 2048;
return(($lastConfig, $lastLustOpts));
}
}
return(("$services|$mdss|$osts|$clts", $thisLustOpts));
}
sub setNames
{
my $type= shift;
my $newNames=shift;
my $oldNames=shift;
my $name;
print "Set Name -- Type: $type Old: $oldNames New: $newNames\n"
if $debug & 8;
# remember, it's ok for names to go away. we just want new ones!
$oldNames=" $oldNames "; # to make pattern match work
foreach $name (split(/\s+/, $newNames))
{
if ($oldNames!~/ $name /)
{
$oldNames.="$name ";
$configChange|=$type;
}
}
$oldNames=~s/^\s+|\s+$//g; # trim leading/trailing space
return($oldNames);
}
# This routine reads partial files AND has /proc specific processing
# code for optimal performance.
sub getProc
{
my $type= shift;
my $proc= shift;
my $tag= shift;
my $ignore=shift;
my $quit= shift;
my $last= shift;
my ($index, $line, $ignoreString);
# matches one or 2 consective //s for pids because when no threads there are 2 of them
logdiag("$proc") if ($utimeMask & 2) && ($proc!~/^\/proc\/?\/\d/) || ($utimeMask & 4) && ($proc=~/^\/proc\/?\/\d/);
if (!open PROC, "<$proc")
{
# but just report it once, but not foe nfs or proc data
logmsg("W", "Couldn't open '$proc'")
if !defined($notOpened{$proc}) && $type!=8 && $type!=13 && $type!=16 && $type!=17;
$notOpened{$proc}=1;
return(0);
}
# Skip beginning if told to do so
$ignore=0 if !defined($ignore);
$quit=(defined($quit)) ? $ignore+$quit : 10000;
for (my $i=0; $i<$ignore; $i++) { ; }
$index=0;
for (my $i=$ignore; $i<$quit; $i++)
{
last if !($line=);
last if defined($last) && $line=~/$last/;
# GENERIC - just prepend tag to records
if ($type==0)
{
$spacer=$tag ne '' ? ' ' : '';
record(2, "$tag$spacer$line");
next;
}
# OST stats
if ($type==1)
{
if ($line=~/^read/) { record(2, "$tag $line"); next; }
if ($line=~/^write/) { record(2, "$tag $line"); next; }
}
# Client RPC and OST brw_stats AND mds/oss disk stats
elsif ($type==2)
{
# for RPC and brw_stats, this block is virtually always 11 entries,
# but the first time an OST is created it's not so we have to stop
# when we hit a blank. In the case of disk stats, we call with
# $last so it quites on the 'totals' row
last if $line=~/^\s+$/;
record(2, "$tag:$index $line");
$index++;
}
# MDS stats
elsif ($type==3)
{
if ($line=~/^mds_/) { record(2, "$tag $line"); next; }
}
# type=4 no longer used
# Memory
elsif ($type==5)
{
next if $line=~/^nr/;
next if $line=~/^numa/;
last if $memOpts!~/[ps]/ && $line=~/^pgre/; # ignore from pgrefill forward
last if $memOpts!~/s/ && $line=~/^pgst/; # ignore from pgstead forward
last if $line=~/^pginode/; # ignore from pginodesteal and below
record(2, "$line");
}
elsif ($type==7)
{
next if ($rawNetIgnore ne '' && $line=~/$rawNetIgnore/) ||
($rawNetFilter ne '' && $line!~/$rawNetFilter/);
record(2, "$tag $line");
next;
}
# NFS
elsif ($type==8)
{
# Can't use type==0 because we don't want a space after $tag
record(2, "$tag$line");
}
# /proc/diskstats & /proc/partitions
# would be nice if we could improve even more since this table can
# get quite large. Note the pattern for cciss MUST match that used
# in formatit.ph!!!
elsif ($type==9)
{
next if $rawDskIgnore ne '' && $line=~/$rawDskIgnore/;
# If disk filter NOT specified in collectl.conf, use the following syntax.
# Even thought it matches internal constant $DiskFilter, it's a little
# faster to as separate if statements
if (!$DiskFilterFlag)
{
if ($line=~/cciss\/c\d+d\d+ /) { record(2, "$tag $line"); next; }
if ($line=~/hd[ab] /) { record(2, "$tag $line"); next; }
if ($line=~/ sd[a-z]+ /) { record(2, "$tag $line"); next; }
if ($line=~/dm-\d+ /) { record(2, "$tag $line"); next; }
if ($line=~/xvd[a-z] /) { record(2, "$tag $line"); next; }
if ($line=~/fio[a-z]+ /) { record(2, "$tag $line"); next; }
if ($line=~/ vd[a-z] /) { record(2, "$tag $line"); next; }
if ($line=~/emcpower[a-z]+ /) { record(2, "$tag $line"); next; }
if ($line=~/psv\d+ /) { record(2, "$tag $line"); next; }
}
else
{
if ($line=~/$DiskFilter/) { record(2, "$tag $line"); next; }
}
}
# /proc/fs/lustre/llite/fsX/stats
elsif ($type==11)
{
if ($line=~/^dirty/) { record(2, "$tag $line"); next; }
if ($line=~/^read_/) { record(2, "$tag $line"); next; }
if ($line=~/^write_/) { record(2, "$tag $line"); next; }
if ($line=~/^open/) { record(2, "$tag $line"); next; }
if ($line=~/^close/) { record(2, "$tag $line"); next; }
if ($line=~/^seek/) { record(2, "$tag $line"); next; }
if ($line=~/^fsync/) { record(2, "$tag $line"); next; }
if ($line=~/^getattr/) { record(2, "$tag $line"); next; }
if ($line=~/^setattr/) { record(2, "$tag $line"); next; }
}
# /proc/fs/lustre/osc/XX/stats
# since I've seen difference instances of SFS report these in different
# locations we have to hunt them out, quitting after 'write' or course.
elsif ($type==12)
{
# This is for the standard CFS/SUN release
if ($sfsVersion eq '')
{
if ($line=~/^read_bytes/) { record(2, "$tag $line"); next; }
if ($line=~/^write_bytes/) { record(2, "$tag $line"); last; }
}
else
{
# and this is the older HP/SFS V2.*
if ($line=~/^ost_read/) { record(2, "$tag $line"); next; }
if ($line=~/^ost_write/) { record(2, "$tag $line"); last; }
}
}
# /proc/*/status - save it all!
elsif ($type==13)
{
# only saving a subset because there is a lot of 'noise' in here
# looks like not exiting early via ^Threads is costing ~10 seconds. If this
# ever turns out to be an issue we could always make collecting the context
# switches optional but for at least now I'm thinking we just do it!
# since ^nonvol is the last entry no need for a test to exit loop earlier
if ($line=~/^Tgid/) { record(2, "$tag $line", undef, 1); next; }
if ($line=~/^Uid/) { record(2, "$tag $line", undef, 1); next; }
if ($line=~/^Vm/) { record(2, "$tag $line", undef, 1); next; }
if ($line=~/^vol/) { record(2, "$tag $line", undef, 1); next; }
if ($line=~/^nonv/) { record(2, "$tag $line", undef, 1); next; }
}
# /proc/slabinfo - only if not doing all of them
elsif ($type==14)
{
$slab=(split(/ /, $line))[0];
record(2, "$tag $line") if defined($slabProc{$slab});
}
# /proc/pid/cmdline - only 1 line long
elsif ($type==16)
{
$line=~s/\000/ /g;
record(2, "$tag $line\n", undef, 1);
last;
}
# identical to type 0, only it writes to process raw file
elsif ($type==17)
{
$spacer=$tag ne '' ? ' ' : '';
record(2, "$tag$spacer$line", undef, 1);
next;
}
# /proc/dev/netstat
elsif ($type==20)
{
record(2, "tcp-$line") if ($tcpFilt=~/I/ && $line=~/^I/) || ($tcpFilt=~/T/ && $line=~/^T/);
}
# /proc/dev/netstat
elsif ($type==21)
{
# no UdpLite or IcmpMsg (at least for now)
if ($line=~/^Icmp:/ && $tcpFilt=~/c/) { record(2, "tcp-$line"); next; }
elsif ($line=~/^Ip/ && $tcpFilt=~/i/) { record(2, "tcp-$line"); next; }
elsif ($line=~/^T/ && $tcpFilt=~/t/) { record(2, "tcp-$line"); next; }
elsif ($line=~/^Udp:/ && $tcpFilt=~/u/) { record(2, "tcp-$line"); next; }
}
# GENERIC 2 - same as generic but support for rawp file
if ($type==99)
{
$spacer=$tag ne '' ? ' ' : '';
record(2, "$tag$spacer$line", undef, 1);
next;
}
}
close PROC;
return(1);
}
# Functionally equivilent to getProc(), but instead has to run a command rather
# than look in proc.
sub getExec
{
my $type= shift;
my $command=shift;
my $tag= shift;
# for now, always send error messages to /dev/null unless we're debugging. This is
# really manditory for perfquery >= ofed 1.5 but let's do it everywhere unless it becomes
# problematic later on.
$command.=' 2>/dev/null' unless $debug & 3;
print "Type: $type Exec: $command\n" if $debug & 2;
# If we can't exec command, only report it once.
if (!open CMD, "$command|")
{
logmsg("W", "Couldn't execute '$command'")
if !defined($notExec[$type]);
$notExec[$type]=1;
return;
}
# Return complete contents of command
my $oneLine='';
if ($type==0)
{
foreach my $line ()
{ record(2, "$tag: $line"); }
}
# Open Fabric
elsif ($type==1)
{
my $lineNum=0;
foreach my $line ()
{
# Skip warnings found in perfquery/ofed 1.5
next if $line=~/^ibwarn/;
# Perfquery V1.5 adds an extra field called CounterSelect2 so ignore.
next if ++$lineNum==13 && ($PQVersion ge '1.5.0');
if ($line=~/^#.*port (\d+)/)
{
# The 0 is a place holder we don't care about, at least not now
$oneLine="$1 0 ";
next;
}
# Since we're not doing anything with hex values this will not include
# the leading 0x, but it will be faster than trying to include it.
$line=~/([0x]*\d+$)/;
$oneLine.="$1 ";
}
$oneLine=~s/ $//;
record(2, "$tag: $oneLine\n");
}
# Voltaire
elsif ($type==2)
{
foreach my $line ()
{
if ($line=~/^PORT=(\d+)$/)
{
$oneLine="$1 ";
next;
}
# If counter, append to list. Note the funky pattern match that will catch
# both decimal and hex numbers.
$oneLine.="$1 " if $line=~/\s(\S*\d)$/;
}
$oneLine=~s/ $//;
record(2, "$tag: $oneLine\n");
}
# impi
elsif ($type==3)
{
foreach my $line ()
{
next if $envFilt ne '' && $line!~/$envFilt/;
record(2, "$tag: $line");
}
}
# just count records
elsif ($type==4)
{
my $count=0;
foreach my $line ()
{ $count++; }
record(2, "$tag: $count\n");
}
close CMD;
}
# This guy is in charge of reading single valued entries, which are
# typical of those found in /sys. The other big difference between
# this and getProc() is it doens't have to deal with all those
# special 'skip', 'ignore', etc flags. Just read the data!
sub getSys
{
my $tag= shift;
my $sys= shift;
my $dir= shift;
my $rawpFlag=shift; # write to rawp file if one and this is defined
my $files= shift;
foreach my $file (@$files)
{
# as of writing this for slub, I'm not expecting file open failures
# but might as well put in here in case needed in the future
my $filespec="$sys/$dir/$file";
if (!open SYS, "<$filespec")
{
# but just report it once
logmsg("E", "Couldn't open '$filespec'")
if !defined($notOpened{$filespec});
$notOpened{$filespec}=1;
return(0);
}
my $line=;
record(2, "$tag $dir $file $line", undef, $rawpFlag);
}
}
sub record
{
my $type= shift;
my $data= shift;
my $recMode= shift; # error recovery mode
my $rawpFlag=shift; # if defined, write to rawp or zrawp
print "$data" if $debug & 4;
# W r i t e T o R A W F i l e
# a few words about writing to the raw gz file... If we fail, we need to
# create a new file and I want to use newLog() since there's a lot going
# one. However, part of newLog() writes the commonHeader as well and that
# in turn calls this routine, so... We pass a flag around indicating we're
# in recovery mode and if writing the common header fails, we have no
# alternative other than to abort.
# when logging raw data to a file $data, the data to write is either an
# interval marker or raw data. Note that when doing plot format to a file
# as well as any terminal based I/O, that all gets handled by dataAnalyze().
if ($logToFileFlag && $rawFlag)
{
if ($zlibFlag)
{
# When flags set, we write 'process' data (identified by '$recFlag1') to a 'rawp'
# file; otherwise just 'raw'
my $rawComp=(defined($rawpFlag) && $recFlag1) ? $ZRAWP : $ZRAW;
$status=$rawComp->gzwrite($data);
if (!$status)
{
$zlibErrors++;
$temp=$recMode ? 'F' : 'E';
logmsg($temp, "Error writing to raw.gz file: ".$rawComp->gzerror());
logmsg("F", "Max Zlib error count exceeded") if $zlibErrors>$MaxZlibErrors;
newLog($filename, "", "", "", "", "", 1);
record(1, sprintf(">>> %.3f <<<\n", $fullTime)) if $recFlag0;
record(1, sprintf(">>> %.3f <<<\n", $fullTime), undef, 1) if $recFlag1;
}
}
else
{
# Same logic as for compressed data above.
my $rawNorm=(defined($rawpFlag) && $recFlag1) ? $RAWP : $RAW;
print $rawNorm $data;
}
}
# G e n e r a t e N u m b e r s F r o m D a t a
# When doing interative reporting OR generating plot data, we need to
# analyze each record as it goes by. This means that in the case of '-P --rawtoo'
# we write to the raw file AND generate the numbers. Also remember that in the
# case of --export we may not end up writing anywhere other than the exported file
dataAnalyze($subsys, $data) if $type==2 && (!$logToFileFlag || $plotFlag || $export ne '');
}
# Design note - this is very subtle, but when creating consecutive files via the log rolling
# mechanism, the last timestamp of one file matches that of the new one. This tells us NOT
# to reset 'last' counters during playback. BUT if newlog() called before new timestamp
# generated, as when $diskChangeFlag set, this does not happen and so you lose 1 interval
# during playback. Not a big deal but worth noting somewhere...
sub newLog
{
my $filename=shift;
my $recDate= shift;
my $recTime= shift;
my $recSecs= shift;
my $recTZ= shift;
my $playback=shift;
my $recMode= shift; # only used during error recovery mode
my ($ss, $mm, $hh, $mday, $mon, $year, $datetime);
my ($dirname, $basename, $command, $fullname, $mode);
my (@disks, $dev, $numDisks, $i, $oldHeader, $oldSubsys, $timesecs, $timezone);
print "NewLog -- Playback: $playback File: $filename Raw: $rawFlag Plot: $plotFlag\n" if $debug & 1;
if ($recDate eq '')
{
# We need EXACT seconds associated with the timestamp of the filename.
# turns out time() and gettimeofday can differ by 5 or 6 msec and when that happens files could end up
# getting rolled 1 second earlier. Therefore always use gettimeofday when using hires to be consistent.
$timesecs=($hiResFlag) ? (Time::HiRes::gettimeofday())[0] : time();
($ss, $mm, $hh, $mday, $mon, $year)=localtime($timesecs);
$datetime=sprintf("%d%02d%02d-%02d%02d%02d",
$year+1900, $mon+1, $mday, $hh, $mm, $ss);
$dateonly=substr($datetime, 0, 8);
$timezone=$LocalTimeZone;
}
else
{
$timesecs=$recSecs;
$datetime="$recDate-$recTime";
$dateonly=$recDate;
$timezone=$recTZ;
}
# Build a common header for ALL files, noting type1 for process
# we only build it if we need it.
$temp="# Date: $datetime Secs: $timesecs TZ: $timezone\n";
$commonHeader= buildCommonHeader(0, $temp);
$commonHeader1=buildCommonHeader(1, $temp) if $recFlag1;
# Now build a slab subheader just to be used for 'raw' and 'slb' files
if ($slubinfoFlag)
{
$slubHeader="#SLUB DATA\n";
foreach my $slab (sort keys %slabdata)
{
# when we have a slab with no aliases, 'first' gets set to that same
# name which in turns ends up on the alias list because it always
# contains 'first' followed by any additional aliases. On the rare
# case we have no alias, which can happen where we have only the root
# slab itself, set the aliases to that slab which will then be skipped.
my $aliaslist=$slabdata{$slab}->{aliaslist};
next if defined($aliaslist) && $slab eq $aliaslist;
$aliaslist=$slab if !defined($aliaslist);
$slubHeader.="#$slab $aliaslist\n";
}
$slubHeader.=sprintf("%s\n", '#'x80);
}
# If generating plot data on terminal, just open everything on STDOUT
# but be SURE set the buffers to flush in case anyone runs as part
# of a script and needs the output immediately.
if ($filename eq "" && $plotFlag)
{
# sigh...
error("Cannot use -P for terminal output of process and 'other' data at the same time")
if $subsys=~/Z/ && length($subsys)>1;
# in the event that someone runs this as a piped command from
# a script and turns off headers things lock up unless these
# files are set to auto-flush.
$zFlag=0;
open $LOG, ">-" or logmsg("F", "Couldn't open LOG for STDOUT"); select $LOG; $|=1;
open BLK, ">-" or logmsg("F", "Couldn't open BLK for STDOUT"); select BLK; $|=1;
open BUD, ">-" or logmsg("F", "Couldn't open BUD for STDOUT"); select BUD; $|=1;
open CLT, ">-" or logmsg("F", "Couldn't open CLT for STDOUT"); select CLT; $|=1;
open CPU, ">-" or logmsg("F", "Couldn't open CPU for STDOUT"); select CPU; $|=1;
open DSK, ">-" or logmsg("F", "Couldn't open DSK for STDOUT"); select DSK; $|=1;
open ELN, ">-" or logmsg("F", "Couldn't open ELN for STDOUT"); select ELN; $|=1;
open ENV, ">-" or logmsg("F", "Couldn't open ENV for STDOUT"); select ENV; $|=1;
open IB, ">-" or logmsg("F", "Couldn't open IB for STDOUT"); select IB; $|=1;
open NFS, ">-" or logmsg("F", "Couldn't open NFS for STDOUT"); select NFS; $|=1;
open NET, ">-" or logmsg("F", "Couldn't open NET for STDOUT"); select NET; $|=1;
open NUMA, ">-" or logmsg("F", "Couldn't open NUMA for STDOUT"); select NUMA; $|=1;
open OST, ">-" or logmsg("F", "Couldn't open OST for STDOUT"); select OST; $|=1;
open TCP, ">-" or logmsg("F", "Couldn't open TCP for STDOUT"); select TCP; $|=1;
open SLB, ">-" or logmsg("F", "Couldn't open SLB for STDOUT"); select SLB; $|=1;
open PRC, ">-" or logmsg("F", "Couldn't open PRC for STDOUT"); select PRC; $|=1;
for (my $i=0; $i<$impNumMods; $i++)
{
open $impGz[$i], ">-" or logmsg("F", "Couldn't open $impKey[$i] for STDOUT"); select $impGz[$i]; $|=1;
}
select STDOUT; $|=1;
return 1;
}
# C r e a t e N e w L o g
# note the way we build files:
# - if name is a dir, the filename starts with hostname.
# - if name not a dir, the filename gets '-host' appended
# - if raw file it also gets date/time but if plot file only date.
$filename= "." if $filename eq ''; # -P and no -f
$filename.=(-d $filename || $filename=~/\/$/) ? "/$Host" : "-$Host";
$filename.=(!$plotFlag || $options=~/u/) ? "-$datetime" : "-$dateonly";
# if the directory doesn't exist (we don't need date/time stamp), create it
$temp=dirname($filename);
if (!-e $temp)
{
logmsg('W', "Creating directory '$temp'");
`mkdir $temp`;
}
# track number of times same file processed, primarily for options 'a/c'. in
# case multiple raw files for same day, only check on initial one
# If we're in playback mode and writing a plotfile, either the user specified
# an option of 'a', 'c' or 'u', we just created it (newFiles{} defined) OR it had
# better not exist! If is does, return it name so a contextual error message
# can be generated.
return $filename if $playback ne "" &&
$options!~/a|c|u/ &&
!defined($newFiles{$filename}) &&
plotFileExists($filename);
# -ou is special in that we're never going to have multiple source files generate
# the same output file so 'a' doesn't mean anything in this context. Furthermore
# if the output file already exists and its update time is less than that of the
# source file, the source file has changed since the output file was created and
# it should and will be overwritten. Finally, the user may also have chosen to
# reprocess a source file with different options and so if 'c' is included the
# file WILL be overwritten even if newer. Whew...
if ($options=~/u/ && plotFileExists($filename))
{
my @files;
@files=glob("$filename*");
my $plotTime=(stat($files[0]))[9];
my $rawTime= (stat($playback))[9];
return($filename) if $plotTime>$rawTime && $options!~/c/;
}
# The only time we force creation of a new file is for the first one of the
# day when in plot create mode (not sure why 'u' too). In all others cases
# we append, which will also create file if not already there.
$newFiles{$filename}++;
if ($options=~/c|u/ && $newFiles{$filename}==1)
{
$mode=">";
$zmode="wb";
}
else
{
$mode=">>";
$zmode="ab";
}
print "NewLog Modes: $mode + $zmode Name: $filename\n" if $debug & 1;
# C r e a t e R A W F i l e
if ($rawFlag)
{
# on subsequent file creates (this is new for V3.1.3) write a terminating time
# stamp, noting this will be the SAME starting timestamp of the new file.
if (!$firstPass)
{
my $fullTime=sprintf("%d.%06d", $intSeconds, $intUsecs);
record(1, sprintf(">>> %.3f <<<\n", $fullTime)) if $recFlag0;
record(1, sprintf(">>> %.3f <<<\n", $fullTime), undef, 1) if $recFlag1;
# Now we can safely close the raw log(s)
closeLogs($subsys, 'r');
}
# In some cases, such as when using --rawtoo (and other situations as well),
# the default filename may only have a datestamp put time back in.
my $rawFilename=$filename;
$rawFilename=~s/$dateonly$/$datetime/;
print "Create raw file: $rawFilename Flag0: $recFlag0 Flag1: $recFlag1\n" if $debug & 8192;
# Unlike plot files, we ALWAYS compress when compression lib exists
$ZRAW=Compress::Zlib::gzopen("$rawFilename.raw.gz", $zmode) or
logmsg("F", "Couldn't open '$rawFilename.raw.gz'") if $zlibFlag && $recFlag0;
$ZRAWP=Compress::Zlib::gzopen("$rawFilename.rawp.gz", $zmode) or
logmsg("F", "Couldn't open '$rawFilename.rawp.gz'") if $zlibFlag && $recFlag1;
open $RAW, "$mode$rawFilename.raw" or
logmsg("F", "Couldn't open '$rawFilename.raw'") if !$zlibFlag && $recFlag0;
open $RAWP, "$mode$rawFilename.rawp" or
logmsg("F", "Couldn't open '$rawFilename.rawp'") if !$zlibFlag && $recFlag1;
# write common header to raw file (record() ignores otherwise). Note that we
# we need to pass along the recovery mode flag because if this record()
# fails it's fatal. we may also need a slub header
record(1, $commonHeader, $recMode) if $recFlag0;
record(1, $commonHeader1, $recMode, 1) if $recFlag1;
record(1, $slubHeader, $recMode, 1) if $slubinfoFlag && $subsys=~/y/i;
# This flag indicated a new file was created and full SLUB records may need to be read
$newRawSlabFlag=1;
}
# C r e a t e P l o t F i l e s
if ($plotFlag)
{
print "Create plot files: $filename.*\n" if $debug & 8192;
# but first close any that might be open
closeLogs($subsys, 'p') if !$firstPass;
# Indicates something needs to be printed
printProcAnalyze($filename) if $procAnalCounter;
printSlabAnalyze($filename) if $slabAnalCounter;
print "Writing file(s): $mode$filename\n" if $msgFlag && !$daemonFlag;
print "Subsys: $subsys\n" if $debug & 1;
# this is already taken care of in playback mode, but when doing -P in
# collection mode we need to clear this since nobody else does!
$headersPrinted=0 if $newFiles{$filename}==1;
# Open 'tab' file in plot mode if processing at least 1 core variable (or extended core)
# OR we're --importing something that prints summary data
$temp="$SubsysCore$SubsysExcore";
if ($subsys=~/[$temp]/ || $impSummaryFlag)
{
$ZLOG=Compress::Zlib::gzopen("$filename.tab.gz", $zmode) or
logmsg("F", "Couldn't open '$filename.tab.gz'") if $zFlag;
open $LOG, "$mode$filename.tab" or
logmsg("F", "Couldn't open '$filename.tab'") if !$zFlag;
$headersPrintedProc=$headersPrintedSlab=0;
}
open BLK, "$mode$filename.blk" or
logmsg("F", "Couldn't open '$filename.blk'") if !$zFlag && $LFlag && $lustOpts=~/D/;
$ZBLK=Compress::Zlib::gzopen("$filename.blk.gz", $zmode) or
logmsg("F", "Couldn't open BLK gzip file") if $zFlag && $LFlag && $lustOpts=~/D/;
open BUD, "$mode$filename.bud" or
logmsg("F", "Couldn't open '$filename.bud'") if !$zFlag && $BFlag;
$ZBUD=Compress::Zlib::gzopen("$filename.bud.gz", $zmode) or
logmsg("F", "Couldn't open BUD gzip file") if $zFlag && $BFlag;
open CPU, "$mode$filename.cpu" or
logmsg("F", "Couldn't open '$filename.cpu'") if !$zFlag && $CFlag;
$ZCPU=Compress::Zlib::gzopen("$filename.cpu.gz", $zmode) or
logmsg("F", "Couldn't open CPU gzip file") if $zFlag && $CFlag;
open CLT, "$mode$filename.clt" or
logmsg("F", "Couldn't open '$filename.clt'") if !$zFlag && $LFlag && $reportCltFlag;
$ZCLT=Compress::Zlib::gzopen("$filename.clt.gz", $zmode) or
logmsg("F", "Couldn't open CLT gzip file") if $zFlag && $LFlag && $reportCltFlag;
# if only doing exceptions, we don't need this file.
if ($options!~/x/)
{
open DSK, "$mode$filename.dsk" or
logmsg("F", "Couldn't open '$filename.dsk'") if !$zFlag && $DFlag;
$ZDSK=Compress::Zlib::gzopen("$filename.dsk.gz", $zmode) or
logmsg("F", "Couldn't open DSK gzip file") if $zFlag && $DFlag;
}
# exception processing for both x and X options
if ($options=~/x/i)
{
open DSKX, "$mode$filename.dskX" or
logmsg("F", "Couldn't open '$filename.dskX'") if !$zFlag && $DFlag;
$ZDSKX=Compress::Zlib::gzopen("$filename.dskX.gz", $zmode) or
logmsg("F", "Couldn't open DSKX gzip file") if $zFlag && $DFlag;
}
if ($XFlag && $NumXRails)
{
open ELN, "$mode$filename.eln" or
logmsg("F", "Couldn't open '$filename.eln'") if !$zFlag;
$ZELN=Compress::Zlib::gzopen("$filename.eln.gz", $zmode) or
logmsg("F", "Couldn't open ELN gzip file") if $zFlag;
}
if ($XFlag && $NumHCAs)
{
open IB, "$mode$filename.ib" or
logmsg("F", "Couldn't open '$filename.ib'") if !$zFlag;
$ZIB=Compress::Zlib::gzopen("$filename.ib.gz", $zmode) or
logmsg("F", "Couldn't open IB gzip file") if $zFlag;
}
open ENV, "$mode$filename.env" or
logmsg("F", "Couldn't open '$filename.env'") if !$zFlag && $EFlag;
$ZENV=Compress::Zlib::gzopen("$filename.env.gz", $zmode) or
logmsg("F", "Couldn't open ENV gzip file") if $zFlag && $EFlag;
open NFS, "$mode$filename.nfs" or
logmsg("F", "Couldn't open '$filename.nfs'") if !$zFlag && $FFlag;
$ZNFS=Compress::Zlib::gzopen("$filename.nfs.gz", $zmode) or
logmsg("F", "Couldn't open NFS gzip file") if $zFlag && $FFlag;
open NUMA, "$mode$filename.numa" or
logmsg("F", "Couldn't open '$filename.numa'") if !$zFlag && $MFlag;
$ZNUMA=Compress::Zlib::gzopen("$filename.numa.gz", $zmode) or
logmsg("F", "Couldn't open NUMA gzip file") if $zFlag && $MFlag;
open NET, "$mode$filename.net" or
logmsg("F", "Couldn't open '$filename.net'") if !$zFlag && $NFlag;
$ZNET=Compress::Zlib::gzopen("$filename.net.gz", $zmode) or
logmsg("F", "Couldn't open NET gzip file") if $zFlag && $NFlag;
open OST, "$mode$filename.ost" or
logmsg("F", "Couldn't open '$filename.ost'") if !$zFlag && $LFlag && $reportOstFlag;
$ZOST=Compress::Zlib::gzopen("$filename.ost.gz", $zmode) or
logmsg("F", "Couldn't open OST gzip file") if $zFlag && $LFlag && $reportOstFlag;
# These next two guys are 'special' because they're not really detail files per se,
# Also note when doing --rawtoo, the data in 'prc' and 'raw' is essentially identical and
# we don't need it on both places. Furthermore, raw is already being compressed.
if (!$rawtooFlag)
{
if (!$procAnalOnlyFlag && $ZFlag)
{
print "Creating PRC file\n" if $debug & 8192;
open PRC, "$mode$filename.prc" or
logmsg("F", "Couldn't open '$filename.prc'") if !$zFlag && $ZFlag;
$ZPRC=Compress::Zlib::gzopen("$filename.prc.gz", $zmode) or
logmsg("F", "Couldn't open PRC gzip file") if $zFlag && $ZFlag;
}
if (!$slabAnalOnlyFlag && $YFlag && !$rawtooFlag)
{
print "Creating SLB file\n" if $debug & 8192;
open SLB, "$mode$filename.slb" or
logmsg("F", "Couldn't open '$filename.slb'") if !$zFlag && $YFlag;
$ZSLB=Compress::Zlib::gzopen("$filename.slb.gz", $zmode) or
logmsg("F", "Couldn't open SLB gzip file") if $zFlag && $YFlag;
}
}
open TCP, "$mode$filename.tcp" or
logmsg("F", "Couldn't open '$filename.tcp'") if !$zFlag && $TFlag;
$ZTCP=Compress::Zlib::gzopen("$filename.tcp.gz", $zmode) or
logmsg("F", "Couldn't open TCP gzip file") if $zFlag && $TFlag;
# Open any detail files associated with --import
for (my $i=0; $i<$impNumMods; $i++)
{
next if $impOpts[$i]!~/d/;
open $impText[$i], "$mode$filename.$impKey[$i]" or
logmsg("F", "Couldn't open '$filename.$impKey[$i]'") if !$zFlag;
$impGz[$i]=Compress::Zlib::gzopen("$filename.$impKey[$i].gz", $zmode) or
logmsg("F", "Couldn't open $impKey[$i] gzip file") if $zFlag;
}
if ($autoFlush)
{
print "Setting non-compressed files to 'autoflush'\n" if $debug & 1;
if (defined($LOG)) { select $LOG; $|=1; }
if (defined(fileno(BLK))) { select BLK; $|=1; }
if (defined(fileno(BUD))) { select BUD; $|=1; }
if (defined(fileno(CLT))) { select CLT; $|=1; }
if (defined(fileno(CPU))) { select CPU; $|=1; }
if (defined(fileno(DSK))) { select DSK; $|=1; }
if (defined(fileno(DSKX))) { select DSKX; $|=1; }
if (defined(fileno(ELN))) { select ELN; $|=1; }
if (defined(fileno(ENV))) { select ENV; $|=1; }
if (defined(fileno(IB))) { select IB; $|=1; }
if (defined(fileno(OST))) { select OST; $|=1; }
if (defined(fileno(NET))) { select NET; $|=1; }
if (defined(fileno(NFS))) { select NFS; $|=1; }
if (defined(fileno(NUMA))) { select NUMA; $|=1; }
if (defined(fileno(PRC))) { select PRC; $|=1; }
if (defined(fileno(SLB))) { select SLB; $|=1; }
if (defined(fileno(TCP))) { select TCP; $|=1; }
select STDOUT; $|=1;
}
}
# P u r g e O l d L o g s
# ... but only if an interval specified
# explicitly purge anything in the logging directory as long it looks like a collectl log
# starting with the host name. in the case of monthly logs, we typically will keep them
# around a LOT longer
if ($purgeDays)
{
my ($day, $mon, $year)=(localtime(time-86400*$purgeDays))[3..5];
my $purgeDate=sprintf("%4d%02d%02d", $year+1900, $mon+1, $day);
$dirname=dirname($filename);
if (opendir(DIR, "$dirname"))
{
while (my $filename=readdir(DIR))
{
next if $filename=~/^\./;
next if $filename=~/log$/;
next if $filename!~/-(\d{8})(-\d{6})*\./ || $1 ge $purgeDate;
unlink "$dirname/$filename";
}
}
else
{
logmsg('E', "Couldn't open '$dirname' for purging");
}
close DIR;
# now do it for the collectl logs themselves, based on the number
# of months, so no days included
($day, $mon, $year)=(localtime(time-86400*$purgeMons*30))[3..5];
$purgeDate=sprintf("%4d%02d", $year+1900, $mon+1);
my $globspec="$dirname/*.log";
foreach my $file (glob($globspec))
{
next if $file!~/-(\d{6})\.log$/;
unlink $file if $1 < $purgeDate;
}
}
# Save as a global for later use. Could probably avoid passing back the name
# on error below, but I'm afraid to change it if I don't have to.
$lastLogPrefix=$filename;
return 1;
}
# Build a common header for ALL files...
sub buildCommonHeader
{
my $rawType= shift;
my $timeZoneInfo=shift;
# if grouping we need to remove subsystems for groups not in
# the associated files
my $tempSubsys=$subsys;
if ($tworawFlag)
{
$tempSubsys=~s/[YZ]//g if $rawType==0;
$tempSubsys=~s/[^YZ]+//g if $rawType==1;
}
# We want to store all the interval(s) being used and not just what
# the user specified with -i. So include i2 if process/slabs and
# i3 more for a placeholder. NOTE - if we're playing back multiple
# files and first doesn't have process data, i2 not set for second
# and so we can't put in header.
$tempInterval=$interval;
$tempInterval.=(defined($interval2)) ? ":$interval2" : ':' if $subsys=~/[yz]/i;
$tempInterval.=($subsys!~/[yz]/i) ? "::$interval3" : ":$interval3"
if $subsys=~/E/;
# For now, these are the only flags I can think of but clearly they
# can grow over time...
my $flags='';
$flags.='d' if $diskChangeFlag;
$flags.='2' if $tworawFlag; # start using 2 instead of 'g'
$flags.='i' if $processIOFlag;
$flags.='s' if $slubinfoFlag;
$flags.='x' if $processCtxFlag;
$flags.='D' if $cpuDisabledFlag;
my $dskNames='';
foreach my $disk (@dskOrder)
{ $dskNames.="$disk "; }
$dskNames=~s/ $//;
my $netNames='';
foreach my $netname (@netOrder)
{ $netNames.=sprintf("$netname:%s ", defined($netSpeeds{$netname}) ? $netSpeeds{$netname} : '??'); }
$netNames=~s/ $//;
my ($sec, $min, $hour, $day, $mon, $year)=localtime($boottime);
my $booted=sprintf "%d%02d%02d-%02d:%02d:%02d", $year+1900, $mon+1, $day, $hour, $min, $sec;
my $commonHeader='';
if ($rawType!=-1 && $playback ne '')
{
$commonHeader.='#'x35;
$commonHeader.=' RECORDED ';
$commonHeader.='#'x35;
$commonHeader.="\n# $recHdr1";
$commonHeader.="\n# $recHdr2" if $recHdr2 ne '';
$commonHeader.="\n";
}
$commonHeader.='#'x80;
$commonHeader.="\n# Collectl: V$Version HiRes: $hiResFlag Options: $cmdSwitches\n";
$commonHeader.="# Host: $Host DaemonOpts: $DaemonOptions\n";
$commonHeader.="# Booted: $boottime [$booted]\n";
$commonHeader.="# Distro: $Distro Platform: $ProductName\n";
$commonHeader.=$timeZoneInfo if defined($timeZoneInfo);
$commonHeader.="# SubSys: $tempSubsys Options: $options Interval: $tempInterval NumCPUs: $NumCpus $Hyper";
$commonHeader.= " CPUsDis: $cpusDisabled" if $cpusDisabled;
$commonHeader.= " NumBud: $NumBud Flags: $flags\n";
$commonHeader.="# Filters: NfsFilt: $nfsFilt EnvFilt: $envFilt TcpFilt: $tcpFilt\n";
$commonHeader.="# HZ: $HZ Arch: $SrcArch PageSize: $PageSize\n";
$commonHeader.="# Cpu: $CpuVendor Speed(MHz): $CpuMHz Cores: $CpuCores Siblings: $CpuSiblings Nodes: $CpuNodes\n";
$commonHeader.="# Kernel: $Kernel Memory: $Memory Swap: $Swap\n";
$commonHeader.="# NumDisks: $dskIndexNext DiskNames: $dskNames\n";
$commonHeader.="# NumNets: $netIndexNext NetNames: $netNames\n";
$commonHeader.="# NumSlabs: $NumSlabs Version: $SlabVersion\n" if $yFlag || $YFlag;
$commonHeader.="# IConnect: NumXRails: $NumXRails XType: $XType XVersion: $XVersion\n" if $NumXRails;
$commonHeader.="# IConnect: NumHCAs: $NumHCAs PortStates: $HCAPortStates IBVersion: $IBVersion PQVersion: $PQVersion\n" if $NumHCAs;
$commonHeader.="# SCSI: $ScsiInfo\n" if $ScsiInfo ne '';
if ($subsys=~/l/i)
{
# Lustre Version and services (if any) info
$commonHeader.="# Lustre: ";
$commonHeader.=" CfsVersion: $cfsVersion" if $cfsVersion ne '';
$commonHeader.=" SfsVersion: $sfsVersion" if $sfsVersion ne '';
$commonHeader.=" LustOpts: $lustOpts Services: $lustreSvcs";
$commonHeader.="\n";
$commonHeader.="# LustreServer: NumMds: $NumMds MdsNames: $MdsNames NumOst: $NumOst OstNames: $OstNames\n"
if $NumOst || $NumMds;
$commonHeader.="# LustreClient: CltInfo: $lustreCltInfo\n"
if $CltFlag && $lustreCltInfo ne ''; # in case all filesystems umounted
# more stuff for Disk Stats
$commonHeader.="# LustreDisks: Num: $NumLusDisks Names: $LusDiskNames\n"
if ($lustOpts=~/D/);
}
for (my $i=0; $i<$impNumMods; $i++) { &{$impUpdateHeader[$i]}(\$commonHeader); }
$commonHeader.="# Comment: $comment\n" if $comment ne '';
$commonHeader.='#'x80;
$commonHeader.="\n";
return($commonHeader);
}
sub writeInterFileMarker
{
# I was torn between putting this test in the one place this routine
# is called or keeping it cleaner and so put it here.
return if $procAnalOnlyFlag;
# for now, only need one for process data
my $marker="# >>> NEW LOG <<<\n";
if ($subsys=~/Z/ && !$rawtooFlag)
{
$ZPRC->gzwrite($marker) or
writeError('prc', $ZPRC) if $zFlag;
print PRC $marker if !$zFlag;
}
}
# see if there is a file that matches this filename root (should't have
# an extension).
sub plotFileExists
{
my $filespec=shift;
my (@files, $file);
@files=glob("$filespec*");
foreach my $file (@files)
{
return(1) if $file!~/raw/;
}
return(0);
}
# In retrospect, there are a number of special cases in here just for playback
# and things might be clearer to do away with this function and move code where
# it applies.
sub setOutputFormat
{
# By default, brief has been initialized to 1 and verbose to 0 but in these
# cases (when not doing --import) we switch to verbose automatically
$verboseFlag=1 if ($subsys ne '' && $subsys!~/^[$BriefSubsys]+$/) || $lustOpts=~/[BDM]/;
$verboseFlag=1 if $memOpts=~/[psPV]/;
$verboseFlag=1 if $tcpFilt eq 'I';
# except as where noted below, columns in verbose mode are assumed different
$sameColsFlag=($verboseFlag) ? 0 : 1;
# Now let's deal with a few special cases where we're in verbose mode but
# the cols are the same after all, such as a single subsystem or '-sCj'
$sameColsFlag=1 if $verboseFlag && (length($subsys)==1 || $subsys=~/^[Cj]+$/);
# Environmental data is multipart if '--envopts M' so we only have same columns when 1 type
$sameColsFlag=0 if $subsys eq 'E' && length($envOpts)>1 && $userEnvOpts=~/M/;
# As usual, lustre complicates things since we can get multiple lines of
# output and if more than 1 clear the flag.
$sameColsFlag=0 if length($lustOpts)>1;
# Finally, if --import modules we've been called at least a second time
if ($impNumMods)
{
# Detail mode forces verbose
if ($impDetailFlag)
{
$verboseFlag=1;
$sameColsFlag=0;
}
# Verbose mode special, because if we don't have any subsystem data and only have 1 type of
# imported data, we still get all data on the same line and won't need to repeat headers every pass.
# On the other hand it we have more than 1 type of data we can't have the same columns
$sameColsFlag=($impSummaryFlag+$impDetailFlag+length($subsys)==1) ? 1 : 0 if $verboseFlag;
# and finally if processing any standard detail data we know we have at least 2 fields, at least
# one of which is our custom import, and so we can't have same columns in effect.
$sameColsFlag=0 if $subsys=~/[A-Z]/; # detail for single -s would have set flag
}
# time doesn't print when not all columns the same AND not something that
# was exported since they's on their own for formatting
if (!$sameColsFlag && $export eq '')
{
$miniDateFlag=$miniTimeFlag=0;
$miniDateTime=$miniFiller='';
}
$briefFlag=($verboseFlag) ? 0 : 1;
print "Set Output -- Subsys: $subsys Verbose: $verboseFlag SameCols: $sameColsFlag\n" if $debug & 1;
# This also feels like a good place to do these
$i1DataFlag=($subsys!~/^[EYZ]+$/i) ? 1 : 0;
$i2DataFlag=($subsys=~/[yYZ]/) ? 1 : 0;
$i3DataFlag=($subsys=~/E/) ? 1 : 0;
}
# Control C Processing
# This will wake us if we're sleeping or let us finish a collection cycle
# if we're not.
sub sigInt
{
print "Ouch!\n" if !$daemonFlag;
$doneFlag=1;
}
sub sigTerm
{
logmsg("W", "Shutting down in response to signal TERM on $myHost...");
$doneFlag=1;
}
sub sigAlrm
{
# This will set next alarm to the next interval that's a multiple of
# our base time. Note the extra 1000usecs which we need as fudge
# Also note that arg[0] always defined with "ALRM" when ualarm below
# fires so we need to use arg[1] as the 'first time' switch for
# logmsg() below.
my ($intSeconds, $intUsecs)=Time::HiRes::gettimeofday();
my $nowUSecs=$intSeconds*1000000+$intUsecs;
my $secs=int(($nowUSecs-$BaseTime+$uAlignInt)/$uAlignInt)*$uAlignInt;
my $waitTime=$BaseTime+$secs-$nowUSecs;
Time::HiRes::ualarm($waitTime+1000);
# message only on the very first call AND when --align since we always
# align on an interval boundary anyway and don't want cluttered messages
logmsg("I", "Waiting $waitTime usecs for time alignment")
if defined($_[1]) && $alignFlag;
# The following is all debug
#($intSeconds, $intUsecs)=Time::HiRes::gettimeofday();
#$nowUSecs2=$intSeconds*1000000+$intUsecs;
#$diff=($nowUSecs2-$nowUSecs)/1000000;
#printf "Start: %f Current: %f Wait: %f Time: %f\n", $BaseTime/1000000, $nowUSecs/1000000, $waitTime/1000000, $diff;
}
# flush buffer(s) on sigUsr1
sub sigUsr1
{
# There should be a small enough number of these to make it worth logging
logmsg("I", "Flushing buffers in response to signal USR1") if !$autoFlush;
logmsg("W", "No need to signal 'USR1' since autoflushing") if $autoFlush;
flushBuffers() if !$autoFlush;
}
sub sigPipe
{
# The only time we're treating a broken pipe as an error is when not in server mode,
# where we simply log but ignore the message since we don't want to quit.
if (!$serverFlag)
{
logmsg("W", "Shutting down due to a broken pipe");
$doneFlag=1;
}
else
{
logmsg("W", "Ignoring broken pipe");
}
}
sub flushBuffers
{
return if !$logToFileFlag;
# Remember, when $rawFlag set we flush everything including process/slab data. But if
# just $rawtooFlag set we those 2 other files aren't open and so we don't flush them.
$flushTime=time+$flush if $flushTime;
logdiag("begin flush") if $utimeMask & 1;
if ($zFlag)
{
if ($rawFlag)
{
# if in raw mode, may be up to 2 buffers to flush
$ZRAW-> gzflush(2)<0 and flushError('raw', $ZRAW) if $recFlag0;
$ZRAWP->gzflush(2)<0 and flushError('raw', $ZRAWP) if $recFlag1;
if (!$plotFlag)
{
logdiag("end flush") if $utimeMask & 1;
return;
}
}
$ZLOG-> gzflush(2)<0 and flushError('log', $ZLOG) if $subsys=~/[a-z]/;
$ZBLK-> gzflush(2)<0 and flushError('blk', $ZBLK) if $LFlag && $lustOpts=~/D/;
$ZBUD-> gzflush(2)<0 and flushError('bud', $ZBUD) if $BFlag;
$ZCPU-> gzflush(2)<0 and flushError('cpu', $ZCPU) if $CFlag;
$ZCLT-> gzflush(2)<0 and flushError('clt', $ZCLT) if $LFlag && $CltFlag;
$ZDSK-> gzflush(2)<0 and flushError('dsk', $ZDSK) if $DFlag && $options!~/x/; # exception only file?
$ZDSKX->gzflush(2)<0 and flushError('dskx',$ZDSKX) if $DFlag && $options=~/x/i;
$ZELN-> gzflush(2)<0 and flushError('eln', $ZELN) if $XFlag && $NumXRails;
$ZIB-> gzflush(2)<0 and flushError('ib', $ZIB) if $XFlag && $NumHCAs;
$ZENV-> gzflush(2)<0 and flushError('env', $ZENV) if $EFlag;
$ZNFS-> gzflush(2)<0 and flushError('nfs', $ZNFS) if $FFlag;
$ZNUMA->gzflush(2)<0 and flushError('net', $ZNET) if $MFlag;
$ZNET-> gzflush(2)<0 and flushError('net', $ZNET) if $NFlag;
$ZOST-> gzflush(2)<0 and flushError('ost', $ZOST) if $LFlag && $OstFlag;
$ZTCP-> gzflush(2)<0 and flushError('tcp', $ZTCP) if $TFlag;
$ZSLB-> gzflush(2)<0 and flushError('slb', $ZSLB) if $YFlag && !$rawtooFlag;
$ZPRC-> gzflush(2)<0 and flushError('prc', $ZPRC) if $ZFlag && !$rawtooFlag;
# handle --import
for (my $i=0; $i<$impNumMods; $i++)
{
# we can only flush detail data if something in buffer or else we'll throw an error!
$impGz[$i]->gzflush(2)<0 and flushError($impKey[$i], $impGz[$i]) if defined($impGz[$i]) && $impDetFlag[$i];
$impDetFlag[$i]=0;
}
}
else
{
if (defined($LOG)) { select $LOG; $|=1; print $LOG ""; $|=0; select STDOUT; }
if (!$plotFlag)
{
logdiag("end flush") if $utimeMask & 1;
return;
}
return if !$plotFlag;
if ($BFlag) { select BUD; $|=1; print BUD ""; $|=0; }
if ($CFlag) { select CPU; $|=1; print CPU ""; $|=0; }
if ($DFlag) { select DSK; $|=1; print DSK ""; $|=0; }
if ($EFlag) { select ENV; $|=1; print ENV ""; $|=0; }
if ($FFlag) { select NFS; $|=1; print NFS ""; $|=0; }
if ($NFlag) { select NET; $|=1; print NET ""; $|=0; }
if ($TFlag) { select TCP; $|=1; print TCP ""; $|=0; }
if ($XFlag && $NumXRails) { select ELN; $|=1; print ELN ""; $|=0; }
if ($XFlag && $NumHCAs) { select IB; $|=1; print IB ""; $|=0; }
if ($YFlag && !$rawtooFlag) { select SLB; $|=1; print SLB ""; $|=0; }
if ($ZFlag && !$rawtooFlag) { select PRC; $|=1; print PRC ""; $|=0; }
if ($LFlag && $CltFlag) { select CLT; $|=1; print CLT ""; $|=0; }
if ($LFlag && $OstFlag) { select OST; $|=1; print OST ""; $|=0; }
if ($LFlag && $lustOpts=~/D/) { select BLK; $|=1; print BLK ""; $|=0; }
# Handle --import
for (my $i=0; $i<$impNumMods; $i++)
{
if (defined($impText[$i])) { select $impText[$i]; $|=1; print {$impText[$i]} ""; $|=0; }
}
if ($options=~/x/i)
{
if ($DFlag) { select DSKX; $|=1; print DSKX ""; $|=0; }
}
select STDOUT;
}
logdiag("end flush") if $utimeMask & 1;
}
sub writeError
{
my $file=shift;
my $desc=shift;
# just print the error and reopen ALL files (since it should be rare)
# we also don't need to set '$recMode' in newLog() since not recursive.
$zlibErrors++;
logmsg("E", "Write error - File: $file Reason: ".$desc->gzerror());
logmsg("F", "Max Zlib error count exceeded") if $zlibErrors>$MaxZlibErrors;
$headersPrinted=0;
newLog($filename, "", "", "", "", "");
}
sub flushError
{
my $file=shift;
my $desc=shift;
# just print the error and reopen ALL files (since it should be rare)
# we also don't need to set '$recMode' in newLog() since not recursive.
$zlibErrors++;
logmsg("E", "Flush error - File: $file Reason: ".$desc->gzerror());
logmsg("F", "Max Zlib error count exceeded") if $zlibErrors>$MaxZlibErrors;
$headersPrinted=0;
newLog($filename, "", "", "", "", "");
}
# write diagnostic record into raw file
sub logdiag
{
my ($intSeconds, $intUsecs)=Time::HiRes::gettimeofday();
my $fullTime=sprintf("%d.%06d", $intSeconds, $intUsecs);
record(1, "### $fullTime $_[0]\n");
}
# Note - ALL errors (both E and F) will be written to syslog. If you want
# others to go there (such as startup/shutdown messages) you need to call
# logsys() directly, but be sure to make sure $filename ne '' (but can't
# unless $filename is known at that point).
sub logmsg
{
my ($severity, $text)=@_;
my ($ss, $mm, $hh, $day, $mon, $year, $msg, $time, $logname, $yymm, $date);
# may need time if in debug and this routine gets called infrequently enough
# that the extra processing is no big deal. Also note that time and gettimeofday
# are not always exactly in sync so when hires loaded, ALWAYS use it
my $timesecs=($hiResFlag) ? (Time::HiRes::gettimeofday())[0] : time();
($ss, $mm, $hh, $day, $mon, $year)=localtime($timesecs);
$time=sprintf("%02d:%02d:%02d", $hh, $mm, $ss);
# always report non-informational messages and if not logging, we're done
# BUT - if not attached to a terminal or not running as a daemon we CAN'T print
# because no terminal to talk to.
# Also, not that we ONLY write to the log when writing to a file and -m
$text="$time $text" if $debug & 1;
print STDERR "$text\n" if $termFlag && !$daemonFlag && ($msgFlag || ($severity eq 'W' && !$quietFlag) || $severity=~/[EF]/ || $debug & 1);
# if we're not writing to a message log always send F/E errors to syslog AND get out if fatal
if (!$msgFlag)
{
logsys($msg, 1) if $severity=~/[EF]/;
exit(1) if $severity eq "F";
}
# Remember: if running as a daemon and NOT -m, we'll never see any messages
# in collectl log OR syslog.
return unless $msgFlag && $filename ne '';
$yymm=sprintf("%d%02d", 1900+$year, $mon+1);
$date=sprintf("%d%02d%02d", 1900+$year, $mon+1, $day);
$msg=sprintf("%s-%s", $severity, $text);
# the log file live in same directory as logs
$logname=(-d $filename) ? $filename : dirname($filename);
$logname.="/$myHost-collectl-$yymm.log";
open MSG, ">>$logname" or logsys("Couldn't open log file '$logname' to write: $msg", 1);
print MSG "$date $time $msg\n" or logsys("Print Error: $! Text: $msg");
close MSG;
logsys($msg) if $severity=~/[EF]/;
exit(1) if $severity=~/F/;
}
sub logsys
{
my $message=shift;
my $force= shift;
# if not writing to a file, only log when forced
return if $PcFlag || ($filename eq '' && !$force);
$x=Sys::Syslog::openlog($Program, "", "user");
$x=Sys::Syslog::syslog("info", "%s", $message);
Sys::Syslog::closelog();
}
# this is for non-fatal messages that are reported before collectl actually
# starts. by saving them, we can then report after the startup message to
# make things cleaner in the log
sub pushmsg
{
my $severity=shift;
my $text= shift;
push @messages, "$severity-$text";
}
sub setFlags
{
my $subsys=shift;
print "SetFlags: $subsys\n" if $debug & 1;
# NOTE - are flags are faster than string compares?
# unfortunately I got stuck using zFlag for ZIP and ZFlag for processes
$bFlag=($subsys=~/b/) ? 1 : 0; $BFlag=($subsys=~/B/) ? 1 : 0;
$cFlag=($subsys=~/c/) ? 1 : 0; $CFlag=($subsys=~/C/) ? 1 : 0;
$dFlag=($subsys=~/d/) ? 1 : 0; $DFlag=($subsys=~/D/) ? 1 : 0;
$EFlag=($subsys=~/E/) ? 1 : 0;
$fFlag=($subsys=~/f/) ? 1 : 0; $FFlag=($subsys=~/F/) ? 1 : 0;
$iFlag=($subsys=~/i/) ? 1 : 0;
$jFlag=($subsys=~/j/) ? 1 : 0; $JFlag=($subsys=~/J/) ? 1 : 0;
$lFlag=($subsys=~/l/) ? 1 : 0; $LFlag=($subsys=~/L/) ? 1 : 0;
$mFlag=($subsys=~/m/) ? 1 : 0; $MFlag=($subsys=~/M/) ? 1 : 0;
$nFlag=($subsys=~/n/) ? 1 : 0; $NFlag=($subsys=~/N/) ? 1 : 0;
$sFlag=($subsys=~/s/) ? 1 : 0;
$tFlag=($subsys=~/t/) ? 1 : 0; $TFlag=($subsys=~/T/) ? 1 : 0;
$xFlag=($subsys=~/x/) ? 1 : 0; $XFlag=($subsys=~/X/) ? 1 : 0;
$yFlag=($subsys=~/y/) ? 1 : 0; $YFlag=($subsys=~/Y/) ? 1 : 0;
$ZFlag=($subsys=~/Z/) ? 1 : 0;
# NOTE - the definition of 'core' as slightly changed and maybe should be
# changed to be 'summary' to better reflect what we're trying to do.
$coreFlag=($subsys=~/[a-z]/) ? 1 : 0;
# by default, all data gets logged in a single file. if the 'tworaw' flag is set,
# we defined flags that control recording into groups based on process/other
$recFlag0=1;
$recFlag1=0;
if ($tworawFlag)
{
$tempSys=$subsys;
$tempSys=~s/[YZ]//g;
$recFlag0=0 if $tempSys eq '';
$recFlag1=1 if $subsys=~/[YZ]/;
}
print "RecFlags: $recFlag0 $recFlag1\n" if $debug & 1 && !$playback;
}
sub setNFSFlags
{
my $nfsFilt=shift;
# Assume no NFS data of any type seen yet. Do it twice to get rid of -w warning
$nfs2CSeen=$nfs3CSeen=$nfs4CSeen=$nfs2SSeen=$nfs3SSeen=$nfs4SSeen=0;
$nfs2CSeen=$nfs3CSeen=$nfs4CSeen=$nfs2SSeen=$nfs3SSeen=$nfs4SSeen=0;
if ($nfsFilt eq '')
{
$nfsCFlag=$nfsSFlag=$nfs2Flag=$nfs3Flag=$nfs4Flag=1;
$nfs2CFlag=$nfs2SFlag=$nfs3CFlag=$nfs3SFlag=$nfs4CFlag=$nfs4SFlag=1;
}
else
{
$nfsCFlag=$nfsSFlag=$nfs2Flag=$nfs3Flag=$nfs4Flag=0;
$nfs2CFlag=$nfs2SFlag=$nfs3CFlag=$nfs3SFlag=$nfs4CFlag=$nfs4SFlag=0;
foreach my $filt (split(/,/, $nfsFilt))
{
# These flags make processing easier/faster later on
if ($filt eq 'c2') { $nfsCFlag=1; $nfs2Flag=1; $nfs2CFlag=1; }
elsif ($filt eq 's2') { $nfsSFlag=1; $nfs2Flag=1; $nfs2SFlag=1; }
elsif ($filt eq 'c3') { $nfsCFlag=1; $nfs3Flag=1; $nfs3CFlag=1; }
elsif ($filt eq 's3') { $nfsSFlag=1; $nfs3Flag=1; $nfs3SFlag=1; }
elsif ($filt eq 'c4') { $nfsCFlag=1; $nfs4Flag=1; $nfs4CFlag=1; }
elsif ($filt eq 's4') { $nfsSFlag=1; $nfs4Flag=1; $nfs4SFlag=1; }
else { error("--nfsfilt option '$filt' not one of 'c2,s2,c3,s3,c4,s4'"); }
}
}
}
sub getSeconds
{
my $date=shift;
my $time=shift;
my ($year, $mon, $day, $hh, $mm, $ss, $seconds);
$year=substr($date, 0, 4);
$mon= substr($date, 4, 2);
$day= substr($date, 6, 2);
$hh= substr($time, 0, 2);
$mm= substr($time, 2, 2);
$ss= substr($time, 4, 2);
return(timelocal($ss, $mm, $hh, $day, $mon-1, $year-1900));
}
# print error and exit if bad datetime without going crazy over all the
# possible purmutations of a bad date/time format
sub checkTime
{
my $switch= shift;
my $datetime=shift;
my $date=0; # can't return ''
my $time=$datetime;
if (length((split(/:/,$datetime))[0])>2)
{
$datetime=~s/^(\d+):?(.*)//;
$date=$1;
$time=$2;
}
$time=($switch eq '--from') ? '00:00:00' : '23:59:59' if $time eq '';
error("Date portion of $switch must be exactly 8 digits")
if $date!=0 && length($date)!=8;
# Make sure time format correct. minimal being HH:MM. supply date and/or ":ss"
$time="0$time" if $time=~/^\d{1}:/;
$time.=":00" if $time!~/^\d{2}:\d{2}:\d{2}$/;
error("$switch time format must be hh:mm[:ss]") if ($time!~/^\d{2}:\d{2}:\d{2}$/);
($hh, $mm, $ss)=split(/:/, $time);
error("$switch specifies invalid time") if ($hh>23 || $mm >59 || $ss>59);
return(($date,"$hh$mm$ss"));
}
sub getDateTime
{
my $seconds=shift;
my ($sec, $min, $hour, $day, $mon, $year)=localtime($seconds);
return(sprintf("%d%02d%02d %02d:%02d:%02d", $year+1900, $mon+1, $day, $hour, $min, $sec));
}
sub checkHiRes
{
if ($TimeHiResCheck && $hiResFlag && -e '/lib/libc.so.6')
{
my $hiResVersion=Time::HiRes->VERSION;
$hiResVersion=~/(\d+)\.(\d+)/;
my $hiResMajor=$1;
my $hiResMinor=$2;
my $glibcAnnounce=`'/lib/libc.so.6'`;
if ($hiResMajor==1 && $hiResMinor<91 &&
$glibcAnnounce=~/GNU C Library stable release version (\d+)\.(\d+)/)
{
my $glibcMajor=$1;
my $glibcMinor=$2;
if ($glibcMajor==2 && ($glibcMinor==4 || $glibcMinor==5))
{
logmsg('W', "WARNING - Your versions of Time::HiRes and glibc are incompatible.");
logmsg('W', " See /opt/hp/collectl/docs/RELEASE-collectl 'Restrictions' for details.");
}
}
}
}
# This is only called during collection, not playback
sub closeLogs
{
my $subsys=shift;
my $ctype= shift;
return if !$logToFileFlag;
# when not specified, close both raw and plot files.
$ctype='rp' if !defined($ctype);
setFlags($subsys);
# C l o s e R a w F i l e ( s )
# closing raw files based on presence of zlib and NOT -oz
if ($rawFlag && $ctype=~/r/)
{
print "Closing raw logs\n" if $debug & 1;
if ($zlibFlag && $logToFileFlag)
{
$ZRAW-> gzclose() if $recFlag0;
$ZRAWP-> gzclose() if $recFlag1;
}
else
{
close $RAW if defined($RAW) && $recFlag0;
close $RAWP if defined($RAWP) && $recFlag1;
}
}
# C l o s e P l o t F i l e ( s )
if ($plotFlag && $ctype=~/p/)
{
print "Closing plot logs\n" if $debug & 1;
# Even if not open, can't hurt to close them.
if (!$zFlag)
{
close LOG;
close BLK;
close BUD;
close CPU;
close CLT;
close DSK;
close DSKX;
close ELN;
close IB;
close ENV;
close NFS;
close NET;
close OST;
close TCP;
close SLB;
close PRC;
}
else # These must be opened in order to close them
{
$temp="$SubsysCore$SubsysExcore";
$ZLOG-> gzclose() if $subsys=~/[$temp]+/ || $impSummaryFlag;
$ZBUD-> gzclose() if $BFlag;
$ZCLT-> gzclose() if $LFlag && CltFlag;
$ZCPU-> gzclose() if $CFlag;
$ZDSK-> gzclose() if $DFlag && $options!~/x/;
$ZDSKX->gzclose() if $DFlag && $options=~/x/i;
$ZELN-> gzclose() if $XFlag && $NumXRails;
$ZIB-> gzclose() if $XFlag && $NumHCAs;
$ZENV-> gzclose() if $EFlag;
$ZNFS-> gzclose() if $FFlag;
$ZNUMA->gzclose() if $MFlag;
$ZNET-> gzclose() if $NFlag;
$ZOST-> gzclose() if $LFlag && $OstFlag;
$ZTCP-> gzclose() if $TFlag;
$ZSLB-> gzclose() if $YFlag && !$rawtooFlag && !$slabAnalOnlyFlag;
$ZPRC-> gzclose() if $ZFlag && !$rawtooFlag && !$procAnalOnlyFlag;
}
# Finally, close any detail logs that may have been opened via --import
for (my $i=0; $i<$impNumMods; $i++)
{
next if $impOpts[$i]!~/d/;
close $impText[$i] if !$zFlag;
$impGz[$i]->gzclose() if $zFlag;
}
}
}
sub loadConfig
{
my $resizePath='';
my ($line, $num, $param, $value, $switches, $file, $openedFlag, $lib);
# If no specified config file, look in /etc and then BinDir and then MyDir
# Note - we can't use ':' as a separator because that screws up windows!
if ($configFile eq '')
{
$configFile="/etc/$ConfigFile;$BinDir/$ConfigFile";
$configFile.=";$MyDir/$ConfigFile" if $BinDir ne '.' && $MyDir ne $BinDir;
}
print "Config File Path: $configFile\n" if $debug & 1;
$openedFlag=0;
foreach $file (split(/;/, $configFile))
{
if (open CONFIG, "<$file")
{
print "Reading Config File: $file\n" if $debug & 1;
$configFile=$file;
$openedFlag=1;
last;
}
}
logmsg("F", "Couldn't open '$configFile'") if !$openedFlag;
$num=0;
foreach $line ()
{
$num++;
next if $line=~/^\s*$|^\#/; # skip blank lines and comments
if ($line!~/=/)
{
logmsg("W", "CONFIG ERROR: Line $num doesn't contain '='. Ignoring...");
next;
}
chomp $line;
($param, $value)=split(/\s*=\s*/, $line);
print "Param: $param Value: $value\n" if $debug & 128;
# S u b s y s t e m s A r e S p e c i a l
# Subsystems -- this is a little tricky because after user overrides
# SubsysCore, SubsysExcore needs to contain all other core subsystems.
if ($param=~/SubsysCore/)
{
# we put everything in 'Excore' and substract what's in 'Core'
$SubsysExcore="$SubsysCore$SubsysExcore";
error("config file entry for '$param' contains invalid subsystem(s) - $value")
if $value!~/^[$SubsysExcore]+$/;
$SubsysCore=$value;
$SubsysExcore=~s/[$SubsysCore]//g;
next;
}
# D a e m o n P r o c e s s i n g
elsif ($param=~/DaemonCommands/ && $daemonFlag)
{
# Pull commmand string off line and add a 'special' end-of-line marker.
# Note that we save off the whole thing for the header and we need the
# ',2' in the split since we can have '=' in the options
$DaemonOptions=$switches=(split(/=\s*/, $line, 2))[1];
$switches.=" -->>>EOL<<<";
# ultimately, we want to prepend these onto the ARG list. The problem is we need to
# preserve the order and the easiest way to do this is to push onto a temp stack
# and pop off when we're done.
my $quote='';
my $switch='';
my @temp;
foreach $param (split(/\s+/, $switches))
{
if ($param=~/^-/)
{
# If new switch, time to write out old one (and arg), but note we're pushing them
# onto a stack so we can retrieve them in the reverse order
if ($switch ne '')
{
push @temp, $switch;
push @temp, $arg if $arg ne '';
}
last if $param eq '-->>>EOL<<<';
$switch=$param;
$arg='';
next;
}
elsif ($quote ne '') # Processing quoted argument
{
$quote='' if $param=~/$quote$/; # this is the last piece
$arg.=" $param";
next if $quote ne '';
}
else # unquoted argument
{
$arg=$param;
$quote=$1 if $param=~/^(['"])/;
}
}
# now put them back, preserving the order
while (my $arg=pop(@temp))
{ unshift(@ARGV, $arg); }
if ($debug & 128)
{ foreach my $arg (@ARGV) { print " $arg "; } print "\n"; }
}
# L i b r a r i e s A r e S p e c i a l T o o
elsif ($param=~/Libraries/)
{
$Libraries=$value;
foreach $lib (split(/\s+/, $Libraries))
{
push @INC, $lib;
}
}
# S t a n d a r d S e t
else
{
$ReqDir=$value if $param=~/^ReqDir/;
$Grep=$value if $param=~/^Grep/;
$Egrep=$value if $param=~/^Egrep/;
$Ps=$value if $param=~/^Ps/;
$Rpm=$value if $param=~/^Rpm/;
$Lspci=$value if $param=~/^Lspci/;
$Lctl=$value if $param=~/^Lctl/;
$resizePath=$value if $param=~/^Resize/;
$ipmitoolPath=$value if $param=~/^Ipmitool/;
$IpmiCache=$value if $param=~/^IpmiCache/;
$IpmiTypes=$value if $param=~/^IpmiTypes/;
# For Infiniband
$PCounter=$value if $param=~/^PCounter/;
$PQuery=$value if $param=~/^PQuery/;
$VStat=$value if $param=~/^VStat/;
$IbDupCheckFlag=$value if $param=~/^IbDupCheckFlag/;
$OfedInfo=$value if $param=~/^OfedInfo/;
$Interval=$value if $param=~/^Interval$/;
$Interval2=$value if $param=~/^Interval2/;
$Interval3=$value if $param=~/^Interval3/;
$LimSVC=$value if $param=~/^LimSVC/;
$LimIOS=$value if $param=~/^LimIOS/;
$LimLusKBS=$value if $param=~/^LimLusKBS/;
$LimLusReints=$value if $param=~/^LimLusReints/;
$LimBool=$value if $param=~/^LimBool/;
$Port=$value if $param=~/^Port/;
$Timeout=$value if $param=~/^Timeout/;
$MaxZlibErrors=$value if $param=~/^ZMaxZlibErrors/;
$LustreSvcLunMax=$value if $param=~/^LustreSvcLunMax/;
$LustreMaxBlkSize=$value if $param=~/^LustreMaxBlkSize/;
$LustreConfigInt=$value if $param=~/^LustreConfigInt/;
$InterConnectInt=$value if $param=~/^InterConnectInt/;
$TermHeight=$value if $param=~/^TermHeight/;
$DefNetSpeed=$value if $param=~/^DefNetSpeed/;
$TimeHiResCheck=$value if $param=~/^TimeHiResCheck/;
$PasswdFile=$value if $param=~/^Passwd/;
$DiskMaxValue=$value if $param=~/^DiskMaxValue/;
$dISKfILTER=$value if $param=~/^DiskFilter/; # note different spelling!!!
$ProcReadTest=$value if $param=~/^ProcReadTest/;
}
}
close CONFIG;
foreach my $bin (split/:/, $resizePath)
{
$Resize=$bin if -e $bin;
}
logmsg('I', "Couldn't find 'resize' so assuming terminal height of 24")
if $Resize eq '';
# Just in case using an older collectl.conf file. Only a problem if
# someone wants to collect IPMI data.
if (!defined($ipmitoolPath))
{
logmsg('E', "Can't find 'Ipmitool' in 'collectl.conf'. Is it old?");
$ipmitoolPath='';
}
# Even though currently one entry, let's make this a path like above
$Ipmitool='';
foreach my $bin (split/:/, $ipmitoolPath)
{
$Ipmitool=$bin if -e $bin;
}
# Unlike other parameters that can be overridden in collectl.conf, we DO need to know if
# that has been done with DiskFilter so we can set the flag correctly and only then
if (defined($dISKfILTER))
{
# the leading/trailing /s are just there for ease of reading in collectl.conf
$DiskFilterFlag=1;
$DiskFilter=$dISKfILTER;
$DiskFilter=~s/^\///;
$DiskFilter=~s/\/$//;
print "DiskFilter set in $configFile: >$DiskFilter<\n" if $debug & 1;
}
}
sub loadSlabs
{
my $slabFilt= shift;
if ($slabinfoFlag)
{
if (!open PROC,")
{
my $slab=(split(/\s+/, $line))[0];
foreach my $filter (split(/,/, $slabFilt))
{
if ($slab=~/^$filter/)
{
$slabProc{$slab}=1;
last;
}
}
}
if ($debug & 1024)
{
print "*** SLABS ***\n";
foreach $slab (sort keys %slabProc)
{ print "$slab\n"; }
}
}
if ($slubinfoFlag)
{
###########################################
# build list of all slabs NOT softlinks
###########################################
opendir SYS, '/sys/slab' or logmsg('F', "Couldn't open '/sys/slab'");
while (my $slab=readdir(SYS))
{
next if $slab=~/^\./;
# If a link, it's actually an alias
$dirname="/sys/slab/$slab";
if (-l $dirname)
{
# If filtering, only keep those aliases that match
next if $slabFilt ne '' && !passSlabFilter($slabFilt, $slab);
# get the name of the slab this link points to
my $linkname=readlink($dirname);
my $rootslab=basename($linkname);
# Note that since scalar returns the number of elements, it's always the index
# we want to write the next entry into. We also want to save a list of the link
# names so we can easily skip over them later.
my $alias=(defined($slabdata{$rootslab}->{aliases})) ? scalar(@{$slabdata{$rootslab}->{aliases}}) : 0;
$slabdata{$rootslab}->{aliases}->[$alias]=$slab;
$slabskip{$slab}=1;
}
else
{
$slabdata{$slab}->{lastobj}=$slabdata{$slab}->{lastslabs}=0;
}
}
##########################################
# secondary filter scan
##########################################
if ($slabFilt ne '')
{
# Note, at this point we only have aliases that pass the filter and so we need
# to keep the entries OR we have entries with no aliases that might still pass
# filters only we couldn't check them yet so we need this second pass.
foreach my $slab (keys %slabdata)
{
delete $slabdata{$slab}
if !defined($slabdata{$slab}->{aliases}) && !passSlabFilter($slabFilt, $slab)
}
}
############################################################
# now find a better name to use, choosing length first
############################################################
# what we want to do here is also build up a list of all the aliases to
# make it easier to insert them into the header as well as display with
# --showslabaliases. Also note is --showrootslabs, we override '$first'
# to that of the slab root name.
foreach my $slab (sort keys %slabdata)
{
my ($first,$kmalloc,$list)=('','',' '); # NOTE - $list set to leading space!
foreach my $alias (@{$slabdata{$slab}->{aliases}})
{
$list.="$alias ";
$kmalloc=$alias if $alias=~/^kmalloc/;
$first=$alias if $alias!~/^kmalloc/ && length($alias)>length($first);
}
$first=$kmalloc if $first eq '';
$first=$slab if $first eq '' || $showRootSlabsFlag;
$slabdata{$slab}->{first}=$first;
$slabfirst{$first}=$slab;
# note that in some cases there is only a single alias in which case 'list' is ''
$list=~s/ $first / /;
$list=~s/^ | $//g;
$slabdata{$slab}->{aliaslist}=$first if $first ne $slab;
$slabdata{$slab}->{aliaslist}.=" $list" if $list ne '';
}
ref($slabfirst); # need to mention it to eliminate -w warning
}
}
sub passSlabFilter
{
my $filters=shift;
my $slab= shift;
foreach my $name (split(/,/, $filters))
{
return(1) if $slab=~/^$name/;
}
return(0);
}
# This needs some explaining... When doing processes, we build a list of all the pids that
# match the --procfilt selection. However, over time a selected command could exist and restart again
# under a different pid and we WANT to pick that up too. So, everytime we check the processes
# and a non-pid selector has been specified we will have to recheck ALL pids to see in any new
# ones show up. Naturally we can skip those in @skipPids and if the flag $pidsOnlyFlag is set
# we can also skip the pid checking. Finally, since over time the list of pids can grow
# unchecked we need to clean out the stale data after every polling cycle.
sub loadPids
{
my $procs=shift;
my ($process, $pid, $ppid, $user, $uid, $cmd, $line, $file, $temp);
my ($type, $value, @ps, $selector, $pidOnly);
# Step 0 - an enhancement! If the process list string is actually a
# filename turn entries into one long string as if entered with --procfilt.
# This makes it possible to have a constant --procfilt parameter yet change
# the process list dynamically, before starting collectl.
if (-e $procs)
{
$temp='';
open TEMP, "<$procs" or logmsg("F", "Couldn't open --procfilt file");
while ($line=)
{
chomp $line;
next if $line=~/^#|^\s*$/; # ignore blank lines
$line=~s/\s+//g; # get rid of ALL whitespace in each line
$temp.="$line," # smoosh it all together
}
$temp=~s/,$//; # get rid of trailing comma
$procs=$temp;
}
# this is pretty brute force, but we're only doing it at startup
# Step 1 - validate list for invalid types OR non-numeric pids
# assume including collectl
$oneThreadFlag=($procs=~/\+/) ? 1 : 0; # handy flag to optimize non-thread cases
$uidMin=$uidMax=$uidSelFlag=0;
foreach $task (split(/,/, $procs))
{
# for now, we don't do too much validation, but be sure to note
# if our pid was requsted via 'p%'
if ($task=~/^([cCpfPuU])\+*(.*)/)
{
$type=$1;
$value=$2;
if ($type=~/u/ && $value=~/(\d+)-(\d+)/)
{
# uids are a special case in that one can specify range or multiple singletons but not multiple
# ranges. when we DO see a range, save it's min/max but DON'T include in the array of selectors
error("you cannot specify multiple uuid ranges in --procfilt") if $uidMin;
$uidMin=$1;
$uidMax=$2;
$uidSelFlag=1;
next;
}
# if we ever do allow this in playback we can't handle 'f'
error("--procfilt f not allowed in playback mode") if $type eq 'f' && $playback ne '';
# pids must be numeric
error("pid $value not numeric in --procfilt") if $type=~/p/i && $value!~/^\d+$/;
# max usernames returned by ps w/o converting to UID looks to be 19
error("cannot use usernames > 19 chars with procfilt") if $type=~/U/ && length($value)>19;
# max command name length returned by ps o comm looks to be 15
error("cannot use commands > 15 chars with procfilt c/C") if $type=~/c/i && length($value)>15;
# when dealing with embedded string in command line, note that spaces
# are converted to NULs, so do it to our match string so it only happens
# once and also be sure to quote any meta charaters the user may have
# in mind to use.
if ($type eq 'f')
{
$task=~s/ /\000/g;
$task=quotemeta($task);
}
push @TaskSelectors, $task;
next;
}
else
{
error("invalid task selection in --procfilt: $task");
}
}
# Step 2 - no longer needed. UIDs loaded earlier
# Step 3 - find pids of all processes that match selection criteria
# be sure to truncate leading spaces since pids are fixed width
# Note: $cmd includes full directory path and args. Furthermore, this is NOT
# what gets stored in /proc/XXX/stat and to make sure we look at the same
# values dynamically as well as staticly, we better pull cmd from the stat
# file itself.
@ps=`ps axo pid,ppid,uid,comm,user`;
my $firstFilePass=1;
foreach $process (@ps)
{
next if $process=~/^\s+PID/;
$process=~s/^\s+//;
chomp $process;
($pid, $ppid, $uid, $cmd, $user)=split(/\s+/, $process);
# if no criteria, select ALL
if ($procs eq '')
{
$pidProc{$pid}=1;
next;
}
# If uid range specified and this UID there, save it noting it's not
# part of the task selection list so we do before the loop below.
if ($uidMin>0 && $uid>=$uidMin && $uid<=$uidMax)
{
$pidOnly=0;
$pidProc{$pid}=1;
next;
}
# select based on criteria, but assume we're not getting a match
$pidOnly=1;
$keepPid=0;
foreach $selector (@TaskSelectors)
{
$pidOnly=0 if $selector!~/^p/;
$uidSelFlag=1 if $selector=~/^u/i; # need to know if doing UID matching
if (($selector=~/^p\+*(.*)/ && $pid eq $1) ||
($selector=~/^P\+*(.*)/ && $ppid eq $1) ||
($selector=~/^c\+*(.*)/ && $cmd=~/$1/) ||
($selector=~/^C\+*(.*)/ && $cmd=~/^$1/) ||
($selector=~/^f\+*(.*)/ && cmdHasString($pid,$1)) ||
($selector=~/^u\+*(.*)/ && $uid eq $1) ||
($selector=~/^U\+*(.*)/ && $user eq $1))
{
# We need to figure out if '+' appended to selector and set flag if so.
# However, since it's extra overhead to maintain %pidThreads, we only set it
# when there are threads to deal with.
$pidThreads{$pid}=(substr($selector, 1, 1) eq '+') ? 1 : 0
if $oneThreadFlag;
$keepPid=1;
last;
}
}
if ($keepPid)
{ $pidProc{$pid}=1; }
else
{ $pidSkip{$pid}=1; }
}
# STEP 4 - deal with threads
# &pidThreads has been set to 1 for any pids we want to watch threads for.
# We clean this up when we clean pids in general. If no pid threads,
# no %pidThreads.
foreach $pid (keys %pidThreads)
{
findThreads($pid) if $pidThreads{$pid};
}
# if a selection list and it's only for pids (and doesn't include uxx-yy), set
# the $pidOnlyFlag so that those are all we ever want to look for
# for force the $pidsOnlyFlag to be set. It's those minor optimization
# in life that count!
$pidOnlyFlag=1 if $procs ne '' && !$uidMin && $pidOnly;
if ($debug & 256)
{
print "PIDS Selected: ";
foreach $pid (sort keys %pidProc)
{
print "$pid ";
}
print "\n";
if ($oneThreadFlag)
{
print "TPIDS Selected: ";
foreach $pid (sort keys %tpidProc)
{
print "$pid ";
}
}
print "\nPIDS Skipped: ";
foreach $pid (sort keys %pidSkip)
{
print "$pid ";
}
print "\n";
print "\$pidOnlyFlag set!!!\n" if $pidOnlyFlag;
}
}
sub loadUids
{
my $passwd=shift;
my (@passwd, $line, $user, $uid);
print "Load UIDS from $passwd\n" if $debug & 1;
if (!-e $passwd)
{
print "WARNING - UID translation file '$passwd' doesn't exist. consider using --passwd\n";
return;
}
@passwd=`$Cat $passwd`;
foreach $line (@passwd)
{
next if $line=~/^\+|^\s*$/; # ignore '+' and blank lines
($user, $uid)=(split(/:/, $line))[0,2];
$UidSelector{$uid}=$user;
}
}
# here we have just found a new pid neither in the list to skip nor to process so
# we have to go back to our selector list and see if it meets the selection specs.
# if so, return the pid AND be sure to add to pidProc{} so we don't come here again.
# There are time we get called and /proc/$pid doesn't exist anymore. This is
# because these are short lived processes that are there where in the directory
# when first read but are gone by the time we want to open them. For efficiency
# we do a test to see if the pid directory exists and then trap later opens in case it
# disappeared by then!
# NOTE - we could probably return 0/1 depending on whether or not pid found, but since
# it's already in the $match variable, we return that for convenience
sub pidNew
{
my $pid=shift;
my ($selector, $type, $param, $match, $cmd, $ppid, $line, $uid);
return(0) if !-e "/proc/$pid/stat";
# if no filter, by defition this is a match
$match=($procFilt ne '') ? 0 : $pid;
# if selecting by uid (either as a range or explict match), try to read this procs
# UID and if not there, no match!
if ($uidSelFlag)
{
$uid=0;
open TMP, ")
{
if ($line=~/^Uid:\s+(\d+)/)
{
$uid=$1;
last;
}
}
# If UID not found it will be 0 and the following always fail
$match=$pid if $uidMin>0 && $uid>=$uidMin && $uid<$uidMax;
}
foreach $selector (@TaskSelectors)
{
$type=substr($selector, 0, 1);
next if $type eq 'p'; # if a pid, can't be a new one
$param=substr($selector, 1);
if ($oneThreadFlag)
{
$param=~s/(\+)//;
$pidThreads{$pid}=($1 eq '+') ? 1 : 0;
}
# match on parents pid? or command?
if ($type=~/[PCc]/)
{
open PROC, ";
($cmd, $ppid)=(split(/ /, $temp))[1,3];
$cmd=~s/[()]//g;
if (($type eq 'P' && $param==$ppid) ||
($type eq 'C' && $cmd=~/^$param/) ||
($type eq 'c' && $cmd=~/$param/))
{
$match=$pid;
last;
}
}
# match on full command path?
elsif ($type=~/f/ && cmdHasString($pid, $param))
{
$match=$pid;
last;
}
# match on UID
elsif ($type=~/u/ && $uid==$param)
{
$match=$pid;
last;
}
# match on username
elsif ($type=~/U/ && defined($UidSelector{$uid}) && $UidSelector{$uid} eq $param)
{
$match=$pid;
last;
}
}
print "%%% Discovered new pid for monitoring: $pid\n"
if $match && ($debug & 256);
$pidProc{$match}=1 if $match!=0;
findThreads($match) if $match && $oneThreadFlag && $pidThreads{$pid};
# since this pid didn't match selection criteria, don't look at it again.
# but, whenever we cycle though all the pids and delete the entire 'skip'
# hash so in case someone we reuses a pid we skipped.
if (!$match)
{
$pidSkip{$pid}=1;
my $num=0;
foreach my $pid (keys %pidSkip)
{ $num++; }
# we only care about the first pid seen at the start of a monitoring interval
if ($firstProcCycle)
{
if ($debug & 4096)
{
my $seconds=int($fullTime);
my $timenow=(split(/\s+/, localtime($seconds)))[3];
printf "$timenow New PID: %5d LastNew: %5d NumPids: %4d NumSkip: $num\n",
$pid, $lastFirstPid, $pid-$lastFirstPid;
}
if ($pid<$lastFirstPid)
{
undef(%pidSkip);
$pid=0; # so we reset $lastFirstPid
print "skipped pids flushed...\n" if $debug & 4096;;
}
$firstProcCycle=0;
$lastFirstPid=$pid;
}
}
return($match);
}
# see if the command that started a process contains a string
sub cmdHasString
{
my $pid= shift;
my $string=shift;
my $line;
# never include ourself when matching by a command line string since it will
# ALWAYS match the collectl command itself
return() if $pid==$$;
# Not an error because proc may have already exited
return(0) if (!open PROC, ";
# since not all processes have command line associated with them be sure to
# check before looking for a match and only return success after making it.
return(!defined($cmdline) || $cmdline!~/$string/ ? 0 : 1)
}
# see if a pid has any active threads
sub findThreads
{
my $pid=shift;
# In some cases the thread owning process may have gone away. When this
# happens we can't open 'task', so act accordingly.
if (!opendir DIR2, "/proc/$pid/task")
{
logmsg("W", "Looks like $pid exited so not looking for new threads");
$pidThreads{$pid}=0;
return;
}
while ($tpid=readdir(DIR2))
{
next if $tpid=~/^\./; # skip . and ..
next if $tpid==$pid; # skip parent beause already covered
# since this routine gets called both at the start when %tpidProc is empty and
# every thread found is new AND during runtime when they may not be, check the
# active thread hash and only include it if not already there
if (!defined($tpidProc{$tpid}))
{
print "%%% Discovered new thread $tpid for pid: $pid\n" if $debug & 256;
$tpidProc{$tpid}=$pid; # add to thread watch list
}
}
}
sub cleanStalePids
{
my ($pid, %pidTemp, %tpidTemp, $removeFlag, $x);
$removeFlag=0;
foreach $pid (keys %pidProc)
{
if ($pidSeen{$pid})
{
$pidTemp{$pid}=1;
# If working with threads, we also need to purge the flag array that tells
# us whether or not to look for thread pids
$tpidTemp{$pid}=$pidThreads{$pid} if $oneThreadFlag;
}
else
{
print "%%% Stale Pid: $pid\n" if $debug & 256;
$removeFlag=1;
}
}
if ($removeFlag)
{
undef %pidProc;
undef %pidThreads;
%pidProc=%pidTemp;
%pidThreads=%tpidTemp;
}
undef %pidTemp;
undef %tpidTemp;
if ($debug & 512)
{
foreach $x (sort keys %pidProc)
{ print "%%% pidProc{}: $x = $pidProc{$x}\n"; }
}
return unless $oneThreadFlag;
# Do it again for threads...
$removeFlag=0;
foreach $pid (keys %tpidProc)
{
if ($tpidSeen{$pid})
{
$pidTemp{$pid}=$tpidProc{$pid};
}
else
{
print "%%% Stale TPid: $pid\n" if $debug & 256;
$removeFlag=1;
}
}
if ($removeFlag)
{
undef %tpidProc;
%tpidProc=%pidTemp;
}
undef %pidTemp;
if ($debug & 512)
{
foreach $x (sort keys %tpidProc)
{ print "%%% tpidProc{}: $x = $tpidProc{$x}\n"; }
}
}
sub showSlabAliases
{
my $slabFilt=shift;
# by setting the slub flag and calling the 'load' routine, we'll get the header
# built
$slubinfoFlag= (-e '/sys/slab') ? 1 : 0;
error("this kernel does not support 'slub-based' slabs") if !$slubinfoFlag;
loadSlabs($slabFilt);
foreach my $slab (sort keys %slabdata)
{
my $aliaslist=$slabdata{$slab}->{aliaslist};
$aliaslist=$slab if !defined($aliaslist);
next if $slab eq $aliaslist;
printf "%-20s %s\n", $slab, $aliaslist if $aliaslist=~/ /;
}
exit(0);
}
sub showVersion
{
$temp='';
$temp.=sprintf("zlib:%s,", Compress::Zlib->VERSION) if $zlibFlag;
$temp.=sprintf("HiRes:%s", Time::HiRes->VERSION) if $hiResFlag;
$temp=~s/,$//;
$version=sprintf("collectl V$Version %s\n\n", $temp ne '' ? "($temp)" : '');
$version.="$Copyright\n";
$version.="$License\n";
printText($version);
exit(0);
}
sub showDefaults
{
printText("Default values by switch:\n");
printText(" Interactive Daemon\n");
printText(" -c -1 -1\n");
printText(" -i 1:$Interval2:$Interval3 $Interval:$Interval2:$Interval3\n");
printText(" --lustsvcs :$LustreConfigInt :$LustreConfigInt\n");
printText(" -s cdn $SubsysCore\n");
printText("Defaults only settable in config file:\n");
printText(" LimSVC = $LimSVC\n");
printText(" LimIOS = $LimIOS\n");
printText(" LimLusKBS = $LimLusKBS\n");
printText(" LimLusReints = $LimLusReints\n");
printText(" LimBool = $LimBool\n");
printText(" Port = $Port\n");
printText(" Timeout = $Timeout\n");
printText(" MaxZlibErrors = $MaxZlibErrors\n");
printText(" Libraries = $Libraries\n") if defined($Libraries);
exit(0);
}
sub envTest
{
$subsys='E';
open ENV, "<$envTestFile" or error("Couldn't open '$envTestFile'");
while (my $line=)
{
next if $line=~/^\s*$|^#/;
dataAnalyze('E', "ipmi $line");
}
close ENV;
$briefFlag=0;
$verboseFlag=1;
intervalPrint(time);
`stty echo` if !$PcFlag && $termFlag && !$backFlag;
}
sub error
{
my $text=shift;
if (defined($text))
{
# when runing as a server, we need to turn off sockFlag otherwise
# printText() will try to send error over socket and we want it local.
$sockFlag=0 if $serverFlag;
`stty echo` if !$PcFlag && $termFlag && !$backFlag;
logmsg("F", "Error: $text") if $daemonFlag;
# we can only call printText() when formatit loaded.
if ($formatitLoaded)
{
printText("Error: $text\n");
printText("type '$Program -h' for help\n");
}
else
{
print "Error: $text\n";
}
exit(1);
}
my $help=<name
--pname name set process name to 'collectl-pname'
-P, --plot generate output in 'plot' format
--procanalyze analyze process data, generating prcs file
--quiet do note echo warning messages on the terminal
-r, --rolllogs time,d,m roll logs at 'time', retaining for 'd' days,
every 'm' minutes [default: d=7,m=1440]
--rawtoo when run with -P, this tell collectl to also
create a raw log file as well
--runas uid[:gui] collectl will change its uid/gid in daemon mode
see man page for details
-R, --runtime duration time to run in format
where unit is w,d,h,m,s
--sep separator specify an alternate plot format separator
--slabanalyze analyze slab data, generating slbs file
--stats same as -oA
-s, --subsys subsys record/playback data from one or more subsystems
--showsubsys for details
--sumstat same as --stats but only summary
--thru time time thru which to playback data (see --from)
--top [type][,num] show top 'num' processes sorted by type
--showtopopts for details
--tworaw synonym for -G and -group, which are now deprecated
--umask mask set output file permissions mask (see man umask)
--utime mask write diagnostic micro timestamps into raw file
--verbose display output in verbose format (automatically
selected when brief doesn't make sense)
-w, --wide print wide field contents (don't use K/M/G)
Synonyms
--utc = -oU
These are Alternate Display Formats
--vmstat show output similar to vmstat
Logging options
--rawtoo used with -P, write raw data to a log as well
--export name[,options] write data to an exported socket/file
Various types of help
-h, --help print this text
-v, --version print version
-V, --showdefs print operational defaults
-x, --helpext extended help
-X, --helpall shows all help concatenated together
--showoptions show all the options
--showsubopts show all substem specific options
--showsubsys show all the subsystems
--showtopopts show --top options
--showheader show file header that 'would be' generated
--showcolheaders show column headers that 'would be' generated
--showslabaliases for SLUB allocator, show non-root aliases
--showrootslabs same as --showslabaliases but use 'root' names
--whatsnew show summary of recent version new features
EOF2
printText("$extended\n");
return if defined($_[0]);
printText("$Copyright\n");
printText("$License\n");
exit(0);
}
sub showSubsys
{
my $subsys=< 1G
G - include decimal point (when it will fit) for numbers > 1G
Exception Reporting
x - report exceptions only (see man page)
X - record all values + exceptions in plot format (see manpage)
Modify results before display (do NOT effect collection)
n - do NOT normalize rates to units/second
Plot File Naming/Creation
a - if plotfile exists, append [default=skip -p file]
c - always create new plot file
u - create unique plot file names - include time
Plot Data Format
1 - plot format with 1 decimal place of precision
2 - plot format with 2 decimal places of precision
z - don't compress output file(s)
File Header Information
i - include file header in output
EOF4
printText($options);
exit(0) if !defined($_[0]);
}
sub showSubopts
{
my $subopts=<[[,],...], and valid types are any
combinations of:
c - any substring in command name
C - command name starts with this string
f - full path of command (including args) contains string
p - pid
P - parent pid
u - any processes owned by this user's UID or in range xxx-yyy
U - any processes owned by this user
NOTE1: if 'procs' is actually a filename, that file will be read and all
lines concatenated together, comma separted, as if typed in as an
argument of --procfilt. Lines beginning with # will be ignored
as comments and blank lines will be skipped.
NOTE2: if any type fields are immediatly followed by a plus sign, any
threads associated with that process will also be reported.
see man page for important restrictions
--procstate Only show processes in one or more of the following states
D - waiting in uninterruptable disk sleep
R - running
S - sleeping in uninterruptable wait
T - traced or stopped
W - paging
Z - zombie
Slab Options and Filters
--slabopts
s - only show slabs with non-zero allocations
S - only show slabs that have changed since last interval
--slabfilt: restricts which slabs are listed, where 'slab's is of the form:
'slab[,slab...]. only slabs whose names start with this name
will be included
TCP Stack Options - these DO effect data collection as well as printing
--tcpfilt
i - ip stats, no brief stats so selecting it alone will force --verbose
t - tcp stats
u - udp stats
c - Icmp Stats
I - ip extended stats
T - tcp extended stats
EOF5
printText($subopts);
exit(0) if !defined($_[0]);
}
sub showTopopts
{
my $subopts=<2;
error("gexpr does not support standard collectl socket I/O via -A") if $sockFlag;
error("host:port must be specified as first parameter") if !defined($hostport) || $hostport eq '';
($gexHost, $gexPort)=split(/:/, $hostport);
error("the port number must be specified") if !defined($gexPort) || $gexPort eq '';
$gexMcastFlag=1 if $gexHost=~/^(\d+)/ && $1>=225 && $1<=239;
error("gexpr subsys options '$gexSubsys' not a proper subset of '$subsys'")
if $subsys ne '' && $gexSubsys ne '' && $gexSubsys!~/^[$subsys]+$/;
$gexColInt=(split(/:/, $interval))[0];
$gexInterval=$gexColInt if $gexInterval eq '';
# convert to the number of samples we want to send
$gexSendCount=int($gexInterval/$gexColInt);
error("gexpr interval of '$gexInterval' is not a multiple of '$gexColInt' seconds")
if $gexColInt*$gexSendCount != $gexInterval;
$gexFlags=$gexMinFlag+$gexMaxFlag+$gexAvgFlag+$gexTotFlag;
error("only 1 of 'min', 'max', 'avg' or 'tot' with 'gexpr'") if $gexFlags>1;
error("'min', 'max', 'avg' & 'tot' require gexpr 'i' that is > collectl's -i")
if $gexFlags && $gexSendCount==1;
if ($gexAlignFlag)
{
my $div1=int(60/$gexColInt);
my $div2=int($gexColInt/60);
error("'align' requires collectl interval be a factor or multiple of 60 seconds")
if ($gexColInt<=60 && $div1*$gexColInt!=60) || ($gexColInt>60 && $div2*60!=$gexColInt);
error("'align' only makes sense when multiple samples/interval") if $gexInterval<=$gexColInt;
error("'lexpr,align' requires -D or --align") if !$gexAlignFlag && !$daemonFlag;
}
# Since gexpr DOES write over a socket but does not use -A, make sure the default
# behavior for -f logs matches that of -A
$rawtooFlag=1 if $filename ne '' && !$plotFlag;
# O p e n S o c k e t
if (!$gexMcastFlag)
{
openSocket($gexHost, $gexPort);
}
else
{
error("must install IO::Socket::Multcast to use multicast feature")
if !eval {require "IO/Socket/Multicast.pm"};
$gexMcast = IO::Socket::Multicast->new() or die "create group";
}
}
sub gexpr
{
# if not time to print and we're not doing min/max/avg/tot, there's nothing to do.
# BUT always make sure time aligns to top of minute based on i=
$gexCounter++;
$gexOutputFlag=(($gexCounter % $gexSendCount) == 0) ? 1 : 0 if !$gexAlignFlag;
$gexOutputFlag=(!(int($lastSecs[$rawPFlag]) % $gexInterval)) ? 1 : 0 if $gexAlignFlag;
return if (!$gexOutputFlag && $gexFlags==0);
if ($gexSubsys=~/c/i)
{
if ($gexSubsys=~/c/)
{
# CPU utilization is a % and we don't want to report fractions
my $i=$NumCpus;
if ($gexGFlag) # for both 'g' OR 'G'
{
sendData('cpu_user', 'percent', $userP[$i]);
sendData('cpu_nice', 'percent', $niceP[$i]);
sendData('cpu_system', 'percent', $sysP[$i]);
sendData('cpu_wio', 'percent', $waitP[$i]);
sendData('cpu_idle', 'percent', $idleP[$i]);
sendData('cpu_num', 'CPUs', $NumCpus);
sendData('proc_total', 'Load/Procs', $loadQue);
sendData('proc_run', 'Load/Procs', $loadRun);
sendData('load_one', 'Load/Procs', $loadAvg1);
sendData('load_five', 'Load/Procs', $loadAvg5);
sendData('load_fifteen', 'Load/Procs', $loadAvg15);
}
if (!$gexGFlag) # if not 'g' use standard collectl names
{
sendData('cputotals.user', 'percent', $userP[$i]);
sendData('cputotals.nice', 'percent', $niceP[$i]);
sendData('cputotals.sys', 'percent', $sysP[$i]);
sendData('cputotals.wait', 'percent', $waitP[$i]);
sendData('cputotals.idle', 'percent', $idleP[$i]);
}
if ($gexGFlag!=1) # 'G' or nothing
{
sendData('cputotals.irq', 'percent', $irqP[$i]);
sendData('cputotals.soft', 'percent', $softP[$i]);
sendData('cputotals.steal','percent', $stealP[$i]);
sendData('ctxint.ctx', 'switches/sec', $ctxt/$intSecs);
sendData('ctxint.int', 'intrpts/sec', $intrpt/$intSecs);
sendData('ctxint.proc', 'pcreates/sec', $proc/$intSecs);
sendData('ctxint.runq', 'runqSize', $loadQue);
}
if (!$gexGFlag) # do it again so that we report ALL cpu %s together
{
sendData('cpuload.avg1', 'loadAvg1', $loadAvg1);
sendData('cpuload.avg5', 'loadAvg5', $loadAvg5);
sendData('cpuload.avg15', 'loadAvg15', $loadAvg15);
}
}
if ($gexSubsys=~/C/)
{
for (my $i=0; $i<$NumCpus; $i++)
{
sendData("cpuinfo.user.cpu$i", 'percent', $userP[$i]);
sendData("cpuinfo.nice.cpu$i", 'percent', $niceP[$i]);
sendData("cpuinfo.sys.cpu$i", 'percent', $sysP[$i]);
sendData("cpuinfo.wait.cpu$i", 'percent', $waitP[$i]);
sendData("cpuinfo.irq.cpu$i", 'percent', $irqP[$i]);
sendData("cpuinfo.soft.cpu$i", 'percent', $softP[$i]);
sendData("cpuinfo.steal.cpu$i", 'percent', $stealP[$i]);
sendData("cpunifo.idle.cpu$i", 'percent', $idleP[$i]);
sendData("cpuinfo.intrpt.cpu$i",'percent', $intrptTot[$i]);
}
}
}
if ($gexSubsys=~/d/i && $gexGFlag!=1)
{
if ($gexSubsys=~/d/)
{
sendData('disktotals.reads', 'reads/sec', $dskReadTot/$intSecs);
sendData('disktotals.readkbs', 'readkbs/sec', $dskReadKBTot/$intSecs);
sendData('disktotals.writes', 'writes/sec', $dskWriteTot/$intSecs);
sendData('disktotals.writekbs', 'writekbs/sec', $dskWriteKBTot/$intSecs);
}
if ($gexSubsys=~/D/)
{
for (my $i=0; $i<@dskOrder; $i++)
{
# preserve display order but skip any disks not seen this interval
$dskName=$dskOrder[$i];
next if !defined($dskSeen[$i]);
next if ($dskFiltKeep eq '' && $dskName=~/$dskFiltIgnore/) || ($dskFiltKeep ne '' && $dskName!~/$dskFiltKeep/);
sendData("diskinfo.reads.$dskName", 'reads/sec', $dskRead[$i]/$intSecs);
sendData("diskinfo.readkbs.$dskName", 'readkbs/sec', $dskReadKB[$i]/$intSecs);
sendData("diskinfo.writes.$dskName", 'writes/sec', $dskWrite[$i]/$intSecs);
sendData("diskinfo.writekbs.$dskName", 'writekbs/sec', $dskWriteKB[$i]/$intSecs);
}
}
}
if ($gexSubsys=~/f/ && $gexGFlag!=1)
{
if ($nfsSFlag)
{
sendData('nfsinfo.SRead', 'SvrReads/sec', $nfsSReadsTot/$intSecs);
sendData('nfsinfo.SWrite', 'SvrWrites/sec', $nfsSWritesTot/$intSecs);
sendData('nfsinfo.Smeta', 'SvrMeta/sec', $nfsSMetaTot/$intSecs);
sendData('nfsinfo.Scommit', 'SvrCommt/sec' , $nfsSCommitTot/$intSecs);
}
if ($nfsCFlag)
{
sendData('nfsinfo.CRead', 'CltReads/sec', $nfsCReadsTot/$intSecs);
sendData('nfsinfo.CWrite', 'CltWrites/sec', $nfsCWritesTot/$intSecs);
sendData('nfsinfo.Cmeta', 'CltMeta/sec', $nfsCMetaTot/$intSecs);
sendData('nfsinfo.Ccommit', 'CltCommt/sec' , $nfsCCommitTot/$intSecs);
}
}
if ($gexSubsys=~/i/ && $gexGFlag!=1)
{
sendData('inodeinfo.dentnum', 'dentrynum', $dentryNum);
sendData('inodeinfo.dentunused', 'dentryunused', $dentryUnused);
sendData('inodeinfo.fhandalloc', 'filesalloc', $filesAlloc);
sendData('inodeinfo.fhandmpct', 'filesmax', $filesMax);
sendData('inodeinfo.inodenum', 'inodeused', $inodeUsed);
}
if ($gexSubsys=~/l/ && $gexGFlag!=1)
{
if ($CltFlag)
{
sendData('lusclt.reads', 'reads/sec', $lustreCltReadTot/$intSecs);
sendData('lusclt.readkbs', 'readkbs/sec', $lustreCltReadKBTot/$intSecs);
sendData('lusclt.writes', 'writes/sec', $lustreCltWriteTot/$intSecs);
sendData('lusclt.writekbs', 'writekbs/sec', $lustreCltWriteKBTot/$intSecs);
sendData('lusclt.numfs', 'filesystems', $NumLustreFS);
}
if ($MdsFlag)
{
my $getattrPlus=$lustreMdsGetattr+$lustreMdsGetattrLock+$lustreMdsGetxattr;
my $setattrPlus=$lustreMdsReintSetattr+$lustreMdsSetxattr;
my $varName=($cfsVersion lt '1.6.5') ? 'reint' : 'unlink';
my $varVal= ($cfsVersion lt '1.6.5') ? $lustreMdsReint : $lustreMdsReintUnlink;
sendData('lusmds.gattrP', 'gattrP/sec', $getattrPlus/$intSecs);
sendData('lusmds.sattrP', 'sattrP/sec', $setattrPlus/$intSecs);
sendData('lusmds.sync', 'sync/sec', $lustreMdsSync/$intSecs);
sendData("lusmds.$varName", "$varName/sec", $varVal/$intSecs);
}
if ($OstFlag)
{
sendData('lusost.reads', 'reads/sec', $lustreReadOpsTot/$intSecs);
sendData('lusost.readkbs', 'readkbs/sec', $lustreReadKBytesTot/$intSecs);
sendData('lusost.writes', 'writes/sec', $lustreWriteOpsTot/$intSecs);
sendData('lusost.writekbs', 'writekbs/sec', $lustreWriteKBytesTot/$intSecs);
}
}
if ($gexSubsys=~/L/ && $gexGFlag!=1)
{
if ($CltFlag)
{
# Either report details by filesystem OR OST
if ($lustOpts!~/O/)
{
for (my $i=0; $i<$NumLustreFS; $i++)
{
sendData("lusost.reads.$lustreCltFS[$i]", 'reads/sec', $lustreCltRead[$i]/$intSecs);
sendData("lusost.readkbs.$lustreCltFS[$i]", 'readkbs/sec', $lustreCltReadKB[$i]/$intSecs);
sendData("lusost.writes.$lustreCltFS[$i]", 'writes/sec', $lustreCltWrite[$i]/$intSecs);
sendData("lusost.writekbs.$lustreCltFS[$i]", 'writekbs/sec', $lustreCltWriteKB[$i]/$intSecs);
}
}
else
{
for (my $i=0; $i<$NumLustreCltOsts; $i++)
{
sendData("lusost.reads.$lustreCltOsts[$i]", 'reads/sec', $lustreCltLunRead[$i]/$intSecs);
sendData("lusost.readkbs.$lustreCltOsts[$i]", 'readkbs/sec', $lustreCltLunReadKB[$i]/$intSecs);
sendData("lusost.writes.$lustreCltOsts[$i]", 'writes/sec', $lustreCltLunWrite[$i]/$intSecs);
sendData("lusost.writekbs.$lustreCltOsts[$i]", 'writekbs/sec', $lustreCltLunWriteKB[$i]/$intSecs);
}
}
}
if ($OstFlag)
{
for ($i=0; $i<$NumOst; $i++)
{
sendData("lusost.reads.$lustreOsts[$i]", 'reads/sec', $lustreReadOps[$i]/$intSecs);
sendData("lusost.readkbs.$lustreOsts[$i]", 'readkbs/sec', $lustreReadKBytes[$i]/$intSecs);
sendData("lusost.writes.$lustreOsts[$i]", 'writes/sec', $lustreWriteOps[$i]/$intSecs);
sendData("lusost.writekbs.$lustreOsts[$i]", 'writekbs/sec', $lustreWriteKBytes[$i]/$intSecs);
}
}
}
if ($gexSubsys=~/m/)
{
if ($gexGFlag) # 'g' or 'G'
{
sendData('mem_total', 'Bytes', $memTot);
sendData('mem_free', 'Bytes', $memFree);
sendData('mem_shared', 'Bytes', $memShared);
sendData('mem_buffers', 'Bytes', $memBuf);
sendData('mem_cached', 'Bytes', $memCached);
sendData('swap_total', 'Bytes', $swapTotal);
sendData('swap_free', 'Bytes', $swapFree);
}
if (!$gexGFlag) # neither
{
sendData('meminfo.tot', 'kb', $memTot);
sendData('meminfo.free', 'kb', $memFree);
sendData('meminfo.shared', 'kb', $memShared);
sendData('meminfo.buf', 'kb', $memBuf);
sendData('meminfo.cached', 'kb', $memCached);
sendData('swapinfo.total', 'kb', $swapTotal);
sendData('swapinfo.free', 'kb', $swapFree);
}
if ($gexGFlag!=1) # nothing or 'G'
{
sendData('meminfo.used', 'kb', $memUsed);
sendData('meminfo.slab', 'kb', $memSlab);
sendData('meminfo.map', 'kb', $memMap);
sendData('meminfo.hugetot', 'kb', $memHugeTot);
sendData('meminfo.hugefree', 'kb', $memHugeFree);
sendData('meminfo.hugersvd', 'kb', $memHugeRsvd);
sendData('swapinfo.used', 'kb', $swapUsed);
sendData('swapinfo.in', 'swaps/sec', $swapin/$intSecs);
sendData('swapinfo.out', 'swaps/sec', $swapout/$intSecs);
sendData('pageinfo.fault', 'faults/sec', $pagefault/$intSecs);
sendData('pageinfo.majfault', 'majflt/sec', $pagemajfault/$intSecs);
sendData('pageinfo.in', 'pages/sec', $pagein/$intSecs);
sendData('pageinfo.out', 'pages/sec', $pageout/$intSecs);
}
}
# gexFlag doesn't apply
if ($gexSubsys=~/M/)
{
for (my $i=0; $i<$CpuNodes; $i++)
{
foreach my $field ('used', 'free', 'slab', 'map', 'anon', 'lock', 'act', 'inact')
{
sendData("numainfo.$field.$i", 'kb', $numaMem[$i]->{$field});
}
}
}
if ($gexSubsys=~/n/i)
{
if ($gexSubsys=~/n/)
{
if ($gexGFlag) # 'g' or 'G'
{
sendData('bytes_in', 'Bytes/sec', $netRxKBTot*1024/$intSecs);
sendData('bytes_out', 'Bytes/sec', $netTxKBTot*1024/$intSecs);
sendData('pkts_in', 'pkts/sec', $netRxPktTot/$intSecs);
sendData('pkts_out', 'pkts/sec', $netTxPktTot/$intSecs);
}
else # neither
{
sendData('nettotals.kbin', 'kb/sec', $netRxKBTot/$intSecs);
sendData('nettotals.pktin', 'pkts/sec', $netRxPktTot/$intSecs);
sendData('nettotals.kbout', 'kb/sec', $netTxKBTot/$intSecs);
sendData('nettotals.pktout', 'pkts/sec', $netTxPktTot/$intSecs);
}
}
if ($gexSubsys=~/N/)
{
for ($i=0; $i<@netOrder; $i++)
{
$netName=$netOrder[$i];
next if !defined($netSeen[$i]);
next if ($netFiltKeep eq '' && $netName=~/$netFiltIgnore/) || ($netFiltKeep ne '' && $netName!~/$netFiltKeep/);
next if $netName=~/lo|sit/;
sendData("nettotals.kbin.$netName", 'kb/sec', $netRxKB[$i]/$intSecs);
sendData("nettotals.pktin.$netName", 'pkts/sec', $netRxPkt[$i]/$intSecs);
sendData("nettotals.kbout.$netName", 'kb/sec', $netTxKB[$i]/$intSecs);
sendData("nettotals.pktout.$netName", 'pkts/sec', $netTxPkt[$i]/$intSecs);
}
}
}
if ($gexSubsys=~/s/ && $gexGFlag!=1)
{
sendData("sockinfo.used", 'sockets', $sockUsed);
sendData("sockinfo.tcp", 'sockets', $sockTcp);
sendData("sockinfo.orphan",'sockets', $sockOrphan);
sendData("sockinfo.tw", 'sockets', $sockTw);
sendData("sockinfo.alloc", 'sockets', $sockAlloc);
sendData("sockinfo.mem", 'sockets', $sockMem);
sendData("sockinfo.udp", 'sockets', $sockUdp);
sendData("sockinfo.raw", 'sockets', $sockRaw);
sendData("sockinfo.frag", 'sockets', $sockFrag);
sendData("sockinfo.fragm", 'sockets', $sockFragM);
}
if ($gexSubsys=~/t/ && $gexGFlag!=1)
{
sendData("tcpinfo.iperrs", 'num/sec', $ipErrors/$intSecs) if $tcpFilt=~/i/;
sendData("tcpinfo.tcperrs", 'num/sec', $tcpErrors/$intSecs) if $tcpFilt=~/t/;
sendData("tcpinfo.udperrs", 'num/sec', $udpErrors/$intSecs) if $tcpFilt=~/u/;
sendData("tcpinfo.icmperrs", 'num/sec', $icmpErrors/$intSecs) if $tcpFilt=~/c/;
sendData("tcpinfo.tcpxerrs", 'num/sec', $tcpExErrors/$intSecs) if $tcpFilt=~/T/;
}
if ($gexSubsys=~/x/i && $gexGFlag!=1)
{
if ($NumXRails)
{
$kbInT= $elanRxKBTot;
$pktInT= $elanRxTot;
$kbOutT= $elanTxKBTot;
$pktOutT=$elanTxTot;
}
if ($NumHCAs)
{
$kbInT= $ibRxKBTot;
$pktInT= $ibRxTot;
$kbOutT= $ibTxKBTot;
$pktOutT=$ibTxTot;
}
sendData("iconnect.kbin", 'kb/sec', $kbInT/$intSecs);
sendData("iconnect.pktin", 'pkt/sec', $pktInT/$intSecs);
sendData("iconnect.kbout", 'kb/sec', $kbOutT/$intSecs);
sendData("iconnect.pktout", 'pkt/sec', $pktOutT/$intSecs);
}
if ($gexSubsys=~/E/i && $gexGFlag!=1)
{
foreach $key (sort keys %$ipmiData)
{
for (my $i=0; $i{$key}}); $i++)
{
my $name=$ipmiData->{$key}->[$i]->{name};
my $inst=($key!~/power/ && $ipmiData->{$key}->[$i]->{inst} ne '-1') ? $ipmiData->{$key}->[$i]->{inst} : '';
sendData("env.$name$inst", $name, $ipmiData->{$key}->[$i]->{value}, '%s');
}
}
}
# if any imported data, it may want to include gexpr output. However this means getting a list of
# 3-tuples to call OUR formatting routines with so the import module doesn't have to.
# NOTE - the assumption is no ganglia specific counters. If there ever are, we'll need to remove
# restriction and ALL imports will have to deal with $gexFlag if called from here
if ($gexGFlag!=1)
{
my (@names, @units, @vals);
for (my $i=0; $i<$impNumMods; $i++) { &{$impPrintExport[$i]}('g', \@names, \@units, \@vals); }
foreach (my $i=0; $i$gexDataMax{$name};
$gexDataTot{$name}+=$value if $gexAvgFlag || $gexTotFlag;
}
return('') if !$gexOutputFlag;
# A c t u a l S e n d H a p p e n s H e r e
# If doing min/max/avg, reset $value
if ($gexFlags)
{
$value=$gexDataMin{$name} if $gexMinFlag;
$value=$gexDataMax{$name} if $gexMaxFlag;
$value=$gexDataTot{$name} if $gexTotFlag;
$value=($gexDataTot{$name}/$gexCounter) if $gexAvgFlag;
}
# Always send send data if not CO mode,but if so only send when it has
# indeed changed OR TTL about to expire
my $valSentFlag=0;
if (!$gexCOFlag || $value!=$gexDataLast{$name} || $gexTTL{$name}==1)
{
$valSentFlag=1;
sendMetaPacket($name, $units);
sendDataPacket($name, $value);
$gexDataLast{$name}=$value;
}
# A fair chunk of work, but worth it
if ($gexDebug & 3)
{
my ($intSeconds, $intUsecs);
if ($hiResFlag)
{
# we have to fully qualify name because or 'require' vs 'use'
($intSeconds, $intUsecs)=Time::HiRes::gettimeofday();
}
else
{
$intSeconds=time;
$intUsecs=0;
}
$intUsecs=sprintf("%06d", $intUsecs);
my ($sec, $min, $hour)=localtime($intSeconds);
my $timestamp=sprintf("%02d:%02d:%02d.%s", $hour, $min, $sec, substr($intUsecs, 0, 3));
printf "$timestamp Name: %-25s Units: %-12s Val: %8d TTL: %d %s\n",
$name, $units, $value, $gexTTL{$name}, ($valSentFlag) ? 'sent' : ''
if $gexDebug & 1 || $valSentFlag;
}
# TTL only applies when in 'CO' mode, noting we already made expiration
# decision above when we saw counter of 1
if ($gexCOFlag)
{
$gexTTL{$name}-- if !$valSentFlag;
$gexTTL{$name}=$gexTTL if $valSentFlag || $gexTTL{$name}==0;
}
}
sub sendMetaPacket
{
my $name= shift;
my $units=shift;
my $string='';
$string.=pack('N', 0x80);
$string.=pack('N', length($myHost));
$string.=packString($myHost);
$string.=pack('N', length($name));
$string.=packString($name);
$string.=pack('N', 0); # spoof
$string.=pack('N', length('double'));
$string.=packString('double');
$string.=pack('N', length($name));
$string.=packString($name);
$string.=pack('N', length($units));
$string.=packString($units);
$string.=pack('N', 3); # slope
$string.=pack('N', 2*$gexTTL*$gexInterval); # time to live
$string.=pack('N', 0); # dmax
$string.=pack('N', 0);
sendUDP($string);
}
sub sendDataPacket
{
my $name= shift;
my $value=shift;
my $string='';
$string.=pack('N', 0x85);
$string.=pack('N', length($myHost));
$string.=packString($myHost);
$string.=pack('N', length($name));
$string.=packString($name);
$string.=pack('N', 0);
$string.=pack('N', 2);
$string.=packString("%s");
$string.=pack('N', length($value));
$string.=packString($value);
sendUDP($string);
}
sub sendUDP
{
my $data=shift;
dumpUDP($data) if $gexDebug & 4;
return if $gexDebug & 8;
my $length=length($data);
for (my $offset=0; $length>0; )
{
# Either send as regular UDP packet(s) OR send to the multicast address
my $bytes=(!$gexMcastFlag) ? send($gexSocket, substr($data, $offset, $gexPktSize), 0, $gexPaddr) :
$gexMcast->mcast_send($data, "$gexHost:$gexPort");
if (!defined($bytes))
{
print "Error: '$!' writing to socket";
last;
}
$offset+=$bytes;
$length-=$bytes;
}
}
sub packString
{
my $string=shift;
my $pad=4-(length($string) % 4);
$pad=0 if $pad==4;
for (my $i=0; $i<$pad; $i++)
{
$string.=pack('c', 0);
}
return($string);
}
sub dumpUDP
{
my $output=shift;
for (my $i=0; $i
Copyright (C)
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License along
with this program; if not, write to the Free Software Foundation, Inc.,
51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
Also add information on how to contact you by electronic and paper mail.
If the program is interactive, make it output a short notice like this
when it starts in an interactive mode:
Gnomovision version 69, Copyright (C) year name of author
Gnomovision comes with ABSOLUTELY NO WARRANTY; for details type `show w'.
This is free software, and you are welcome to redistribute it
under certain conditions; type `show c' for details.
The hypothetical commands `show w' and `show c' should show the appropriate
parts of the General Public License. Of course, the commands you use may
be called something other than `show w' and `show c'; they could even be
mouse-clicks or menu items--whatever suits your program.
You should also get your employer (if you work as a programmer) or your
school, if any, to sign a "copyright disclaimer" for the program, if
necessary. Here is a sample; alter the names:
Yoyodyne, Inc., hereby disclaims all copyright interest in the program
`Gnomovision' (which makes passes at compilers) written by James Hacker.
, 1 April 1989
Ty Coon, President of Vice
This General Public License does not permit incorporating your program into
proprietary programs. If your program is a subroutine library, you may
consider it more useful to permit linking proprietary applications with the
library. If this is what you want to do, use the GNU Lesser General
Public License instead of this License.
collectl-3.6.9/README-WINDOWS 0000644 0001750 0001750 00000000723 12230241726 013507 0 ustar mjs mjs collectl only runs in playback mode on windows so you
will have to copy over one or more raw files to use it.
You will also have to have perl installed.
To Install
- unpack src tarball, which can be done by most utilitis, into a
temporary directory
- create a directory to install collectl into, say \collectl
- copy the following files into \collectl
- collectl
- collectl.conf
- *ph
You're done! To verify the installation run
\collectl\collectl.pl -v
collectl-3.6.9/nvidia.ph 0000644 0001750 0001750 00000032536 12230241726 013311 0 ustar mjs mjs # copyright, 2003-2009 Hewlett-Packard Development Company, LP
# NOTE - by default, this module only collects data once a minute, which you can change
# with the i=x parameter (eg --import nvidia,i=x). Regardless of the collection interval,
# date will be reported every interval in brief/verbose formats to provide a consisent
# set of output each monitoring cycle.
# n v d i a m o n i t o r i n g
use strict;
# Allow reference to collectl variables, but be CAREFUL as these should be treated as readonly
our ($Version, $miniFiller, $rate, $SEP, $datetime, $miniInstances, $interval, $showColFlag, $playback);
my ($nvidiaInterval, $nvidiaImportCount, $nvidiaSampleCounter, $nvidiaAllFlag);
my ($nvidiaOpts, $nvidiaNumGpus,$nvidiaProdName, $nvidiaDriverVer);
my (@nvidiaTemp, @nvidiaFan, @nvidiaGpuUtil, @nvidiaMemUtil);
my ($nvidiaTempTot, $nvidiaFanTot, $nvidiaGpuUtilTot, $nvidiaMemUtilTot);
my ($nvidiaTempTOT, $nvidiaFanTOT, $nvidiaGpuUtilTOT, $nvidiaMemUtilTOT);
my $nvidiaGpuFirst;
my $nvidiaGpuNum=-1;
my $nvidiaSection=0;
my $nvidiaLastValFlag;
my $nvidiaCommand='/usr/bin/nvidia-smi';
my $nvidiaFullCmd="$nvidiaCommand -q -a";
sub nvidiaInit
{
my $impOptsref=shift;
my $impKeyref= shift;
# If we ever run with a ':' in the interval, we need to be sure we're
# only looking at the main one.
my $nvidiaInterval1=(split(/:/, $interval))[0];
# For now, only options are a, 'i=, d, s and sd
$nvidiaOpts='';
$nvidiaInterval=60;
if (defined($$impOptsref))
{
foreach my $option (split(/,/,$$impOptsref))
{
my ($name, $value)=split(/=/, $option);
error("invalid misc option: '$name'") if $name ne 'a' && $name ne 'i' && $name!~/^[ds]+$/;
$nvidiaInterval=$value if $name eq 'i';
$nvidiaAllFlag=1 if $name eq 'a';
$nvidiaOpts=$name if $name=~/^[ds]+$/;
}
}
# Special case: if --showcowheaders, we're called with -i0 and need to reset so the
# division below doesn't bomb out.
$nvidiaInterval1=1 if $showColFlag;
$nvidiaImportCount=int($nvidiaInterval/$nvidiaInterval1);
error("nvidia interval option not a multiple of '$nvidiaInterval1' seconds")
if $nvidiaInterval1*$nvidiaImportCount != $nvidiaInterval;
$nvidiaOpts='s' if $nvidiaOpts eq '';
$$impOptsref=$nvidiaOpts;
$$impKeyref='nvid';
# since totals only initialized in 'analyze' routine, do it here too in case no data
# in raw file during playback.
$nvidiaTempTot=$nvidiaFanTot=$nvidiaGpuUtilTot=$nvidiaMemUtilTot=0;
# R u n C o m m a n d I n R e a l t i m e M o d e
# when running in collection mode only, we need to make sure we're nvidia enabled
if ($playback eq '')
{
if ($playback eq '' && !-e $nvidiaCommand || $Version lt '3.5.0-3')
{
my $message="cannot find '$nvidiaCommand'" if !-e $nvidiaCommand;
$message="this nvidia.ph requires at least collectl V3.5.1" if $Version lt '3.5.1';
pushmsg('W', "$message. GPU monitoring disabled");
$$impOptsref=$$impKeyref='';
return(-1);
}
# Now run command once to get a get a few constants for header AND number of GPUs
# NOTE: if it returns its names its probably a help message and the desired commands
# are not suppoted, hence not a GPU we can monitor.
my $output=`$nvidiaFullCmd`;
if ($output=~/nvidia-smi/)
{
pushmsg('W', "nvidia GPU doesn' support '$nvidiaFullCmd'. GPU monitoring disabled");
$$impOptsref=$$impKeyref='';
return(-1);
}
$nvidiaNumGpus=0;
foreach my $line ((split(/\n/, $output)))
{
chomp $line;
$nvidiaDriverVer=$1 if $line=~/Driver Version\s+:\s+(.*)/;
if ($line=~/Product Name\s+:\s+(.*)/)
{
$nvidiaProdName=$1;
$nvidiaNumGpus++;
}
}
$nvidiaSampleCounter=-1;
}
return(1);
}
# Nothing to add to header
sub nvidiaUpdateHeader
{
my $lineref=shift;
$$lineref.="# GPU: Type: nvidia $nvidiaProdName Version: $nvidiaDriverVer NumGPU: $nvidiaNumGpus\n";
}
sub nvidiaGetHeader
{
my $headerref=shift;
# Make sure playback file contains nvidia gpu data and if not, just set number
# of gpus to 1 to make display logic work because maybe this is intended.
if ($$headerref!~/GPU:(.*)Version/ || $1!~/nvidia/)
{
logmsg('W', "playback file does not contain nvidia data");
$nvidiaNumGpus=1;
}
else
{
# Now parse header for real
$$headerref=~/GPU:.*Type: (.*).*Version: (.*)\s+NumGPU: (\d+)/;
$nvidiaDriverVer=$2;
$nvidiaNumGpus=$3;
}
}
sub nvidiaGetData
{
# only read counters when OUR interval directs us to.
return if ($nvidiaSampleCounter++ % $nvidiaImportCount)!=0;
getExec(0, $nvidiaFullCmd, 'nvid');
}
sub nvidiaInitInterval
{
# all initialization totals MUST be done in Analyze() because this not called
# each pass when writing to a raw file.
}
sub nvidiaAnalyze
{
my $type= shift;
my $dataref=shift;
# we need to know when last counter seen AND we can't use InitInterval()
$nvidiaLastValFlag=0;
# This is a mess! Difference drivers generate different output
if ($nvidiaDriverVer lt '270.41.19') # until I know better
{
# GPU number always preceeds data
$nvidiaGpuNum=$1 if $$dataref=~/^GPU (\d+)/;
my $i=$nvidiaGpuNum;
$nvidiaTemp[$i]=$1 if $$dataref=~/^Temp.*?(\d+)/;
$nvidiaFan[$i]=$1 if $$dataref=~/^Fan.*?(\d+)/;
$nvidiaGpuUtil[$i]=$1 if $$dataref=~/^GPU.*: (\d+)/;
# This indicates the end of a set of data. If more data elements
# are ever added, we'll need a different 'last' variable test
if ($$dataref=~/^Memory.*?(\d+)/)
{
$nvidiaMemUtil[$i]=$1;
$nvidiaLastValFlag=1;
}
}
elsif ($nvidiaDriverVer ge '270.41.19')
{
# for this driver version this isn't a number but a string so just bump index
# and make a note of it so we recognize the start of a new interval
if ($$dataref=~/^GPU (.*)/ && $$dataref!~/UUID/)
{
$nvidiaGpuFirst=$1 if !defined($nvidiaGpuFirst);
$nvidiaGpuNum=-1 if $1 eq $nvidiaGpuFirst;
$nvidiaGpuNum++;
}
$nvidiaSection=10 if $$dataref=~/^PCI/;
$nvidiaSection=20 if $$dataref=~/^Fan/;
$nvidiaSection=30 if $$dataref=~/^Memory Usage/;
$nvidiaSection=40 if $$dataref=~/^Utilization/;
$nvidiaSection=50 if $$dataref=~/^Ecc Mode/;
$nvidiaSection=60 if $$dataref=~/^ECC Errors/;
$nvidiaSection=70 if $$dataref=~/^Temperature/;
$nvidiaSection=80 if $$dataref=~/^Power Readings/;
$nvidiaSection=90 if $$dataref=~/^Clocks/;
# for now we only care about Fan, Utiliztion and Temp
return if $nvidiaSection!=20 && $nvidiaSection!=40 && $nvidiaSection!=70;
my $i=$nvidiaGpuNum;
$nvidiaFan[$i]=$1 if $nvidiaSection==20 && $$dataref=~/^Fan.*?(\d+)/;
$nvidiaGpuUtil[$i]=$1 if $nvidiaSection==40 && $$dataref=~/^Gpu.*: (\d+)/;
$nvidiaMemUtil[$i]=$1 if $nvidiaSection==40 && $$dataref=~/^Memory.*?(\d+)/;
if ($nvidiaSection==70 && $$dataref=~/^Gpu.*?(\d+)/)
{
$nvidiaTemp[$i]=$1;
$nvidiaLastValFlag=1;
}
}
# We can only do initialization and update totals when we see last data element (actually
# we could have updated the totals as each element seen but this is a little easier.
if ($nvidiaLastValFlag)
{
my $i=$nvidiaGpuNum;
# These only done once/interval, not per GPU
$nvidiaTempTot=$nvidiaFanTot=$nvidiaGpuUtilTot=$nvidiaMemUtilTot=0 if $i==0;
$nvidiaTempTot+= $nvidiaTemp[$i];
$nvidiaFanTot+= $nvidiaFan[$i];
$nvidiaGpuUtilTot+=$nvidiaGpuUtil[$i];
$nvidiaMemUtilTot+=$nvidiaMemUtil[$i];
}
}
sub nvidiaPrintBrief
{
my $type=shift;
my $lineref=shift;
if ($type==1) # header line 1
{
$$lineref.="<------nvidia------->";
}
elsif ($type==2) # header line 2
{
$$lineref.=" Temp Fan% Gpu% Mem% ";
}
elsif ($type==3) # data
{
my $num=$nvidiaNumGpus;
$$lineref.=sprintf(" %4d %4d %4d %4d ",
$nvidiaTempTot/$num, $nvidiaFanTot/$num, $nvidiaGpuUtilTot/$num, $nvidiaMemUtilTot/$num);
}
elsif ($type==4) # reset 'total' counters
{
$nvidiaTempTOT=$nvidiaFanTOT=$nvidiaGpuUtilTOT=$nvidiaMemUtilTOT=0;
}
elsif ($type==5) # increment 'total' counters
{
$nvidiaTempTOT+= $nvidiaTempTot;
$nvidiaFanTOT+= $nvidiaFanTot;
$nvidiaGpuUtilTOT+=$nvidiaGpuUtilTot;
$nvidiaMemUtilTOT+=$nvidiaMemUtilTot;
}
elsif ($type==6) # print 'total' counters
{
my $num=$nvidiaNumGpus;
printf(" %4d %4d %4d %4d ",
int($nvidiaTempTOT/$num/$miniInstances), int($nvidiaFanTOT/$num/$miniInstances),
int($nvidiaGpuUtilTOT/$num/$miniInstances), int($nvidiaMemUtilTOT/$num/$miniInstances));
}
}
sub nvidiaPrintVerbose
{
my $printHeader=shift;
my $homeFlag= shift;
my $lineref= shift;
if ($nvidiaOpts=~/s/)
{
my $line='';
if ($printHeader)
{
$line.="\n" if !$homeFlag;
$line.="# GPU STATISTICS\n";
$line.="#$miniFiller Temp Fan% GPU% Mem%\n";
}
my $num=$nvidiaNumGpus;
$$lineref=$line;
$$lineref.=sprintf("$datetime %3d %3d %3d %3d\n",
int($nvidiaTempTot/$num), int($nvidiaFanTot/$num),
int($nvidiaGpuUtilTot/$num), int($nvidiaMemUtilTot/$num));
}
if ($nvidiaOpts=~/d/)
{
my $line='';
if ($printHeader)
{
$line.="\n" if !$homeFlag;
$line.="# GPU DETAIL\n";
$line.="#$miniFiller Num Temp Fan% GPU% Mem%\n";
}
$$lineref=$line;
for (my $i=0; $i<$nvidiaNumGpus; $i++)
{
$$lineref.=sprintf("$datetime %3d %3d %3d %3d %3d\n",
$i, $nvidiaTemp[$i], $nvidiaFan[$i], $nvidiaGpuUtil[$i], $nvidiaMemUtil[$i])
}
}
}
sub nvidiaPrintPlot
{
my $type= shift;
my $ref1= shift;
# H e a d e r s
# Summary
if ($type==1 && $nvidiaOpts=~/s/)
{
$$ref1.="[NVIDIA]Temp${SEP}[NVIDIA]Fan${SEP}[NVIDIA]Gpu${SEP}[NVIDIA]Mem${SEP}"
}
# Detail
if ($type==2 && $nvidiaOpts=~/d/)
{
for (my $i=0; $i<$nvidiaNumGpus; $i++)
{
$$ref1.="[NVIDIA:$i]Temp${SEP}[NVIDIA:$i]Fan${SEP}[NVIDIA:$i]Gpu${SEP}[NVIDIA:$i]Mem${SEP}"
}
}
# D a t a
# Summary
if ($type==3 && $nvidiaOpts=~/s/)
{
my $num=$nvidiaNumGpus;
$$ref1.=sprintf("$SEP%d$SEP%d$SEP%d$SEP%d",
$nvidiaTempTot/$num, $nvidiaFanTot/$num, $nvidiaGpuUtilTot/$num, $nvidiaMemUtilTot/$num)
}
# Detail
if ($type==4 && $nvidiaOpts=~/d/)
{
for (my $i=0; $i<$nvidiaNumGpus; $i++)
{
$$ref1.=sprintf("$SEP%d$SEP%d$SEP%d$SEP%d",
$nvidiaTemp[$i], $nvidiaFan[$i], $nvidiaGpuUtil[$i], $nvidiaMemUtil[$i]);
}
}
}
sub nvidiaPrintExport
{
my $type= shift;
my $ref1= shift;
my $ref2= shift;
my $ref3= shift;
# counters are only returned based on "i=" or default of 60 seconds
return if !$nvidiaAllFlag && (($nvidiaSampleCounter-1) % $nvidiaImportCount)!=0;
if ($nvidiaOpts=~/s/)
{
if ($type eq 'l')
{
push @$ref1, "nvidia.temp"; push @$ref2, sprintf("%d", $nvidiaTempTot/$nvidiaNumGpus);
push @$ref1, "nvidia.fan"; push @$ref2, sprintf("%d", $nvidiaFanTot/$nvidiaNumGpus);
push @$ref1, "nvidia.gpu"; push @$ref2, sprintf("%d", $nvidiaGpuUtilTot/$nvidiaNumGpus);
push @$ref1, "nvidia.mem"; push @$ref2, sprintf("%d", $nvidiaMemUtilTot/$nvidiaNumGpus);
}
elsif ($type eq 's')
{
my $num=$nvidiaNumGpus;
$$ref1.=sprintf("(nvidiatot (temp %d)(fan %d)(gpu %d)(mem %d))\n",
$nvidiaTempTot/$num, $nvidiaFanTot/$num, $nvidiaGpuUtilTot/$num, $nvidiaMemUtilTot/$num);
}
elsif ($type eq 'g')
{
push @$ref2, 'pct', 'pct', 'pct', 'pct';
push @$ref1, "nvidia.temp"; push @$ref3, sprintf("%d", $nvidiaTempTot/$nvidiaNumGpus);
push @$ref1, "nvidia.fan"; push @$ref3, sprintf("%d", $nvidiaFanTot/$nvidiaNumGpus);
push @$ref1, "nvidia.gpu"; push @$ref3, sprintf("%d", $nvidiaGpuUtilTot/$nvidiaNumGpus);
push @$ref1, "nvidia.mem"; push @$ref3, sprintf("%d", $nvidiaMemUtilTot/$nvidiaNumGpus);
}
}
if ($nvidiaOpts=~/d/)
{
if ($type=~/[gl]/)
{
for (my $i=0; $i<$nvidiaNumGpus; $i++)
{
if ($type eq 'l')
{
push @$ref1, "nvidia.temp$i"; push @$ref2, sprintf("%d", $nvidiaTemp[$i]);
push @$ref1, "nvidia.fan$i"; push @$ref2, sprintf("%d", $nvidiaFan[$i]);
push @$ref1, "nvidia.gpu$i"; push @$ref2, sprintf("%d", $nvidiaGpuUtil[$i]);
push @$ref1, "nvidia.mem$i"; push @$ref2, sprintf("%d", $nvidiaMemUtil[$i]);
}
elsif ($type eq 'g')
{
push @$ref2, 'pct', 'pct', 'pct', 'pct';
push @$ref1, "nvidia.temp$i"; push @$ref3, sprintf("%d", $nvidiaTemp[$i]);
push @$ref1, "nvidia.fan$i"; push @$ref3, sprintf("%d", $nvidiaFan[$i]);
push @$ref1, "nvidia.gpu$i"; push @$ref3, sprintf("%d", $nvidiaGpuUtil[$i]);
push @$ref1, "nvidia.mem$i"; push @$ref3, sprintf("%d", $nvidiaMemUtil[$i]);
}
}
}
elsif ($type eq 's')
{
$$ref2.="(nvidiainfo\n";
$$ref2.=" (num";
for (my $i=0; $i<$nvidiaNumGpus; $i++)
{ $$ref2.=" $i"; }
$$ref2.=")\n";
my ($temp, $fan, $gpu, $mem)=('','','','');
for (my $i=0; $i<$nvidiaNumGpus; $i++)
{
$temp.=" $nvidiaTemp[$i]";
$fan.= " $nvidiaFan[$i]";
$gpu.= " $nvidiaGpuUtil[$i]";
$mem.= " $nvidiaMemUtil[$i]";
}
$$ref2.=" (temp$temp)\n";
$$ref2.=" (fan$fan)\n";
$$ref2.=" (gpu$gpu)\n";
$$ref2.=" (mem$mem))\n";
}
}
}
1;
collectl-3.6.9/sexpr.ph 0000644 0001750 0001750 00000074235 12230241726 013202 0 ustar mjs mjs # copyright, 2003-2009 Hewlett-Packard Development Company, LP
# Debug
# 1 - show useful stuff
# 2 - show creation of header file
#####################################################
# S - E x p r e s s i o n S u p p o r t
#####################################################
my ($sexprDebug, $sexprFilename, $sexprDispatch, $sexprHeaderPrinted);
sub sexpr
{
sexprRaw() if $sexprDispatch==1;
sexprRate() if $sexprDispatch==2;
}
sub sexprInit
{
error('--showcolheader not supported by sexpr') if $showColFlag;
$sexprDebug=0;
$sexprDispatch=0;
$sexprFilename='';
$sexprSubsys=$subsys;
$sexprHeaderPrinted=($sockFlag) ? 1 : 0; # we never write local snapshot in socket mode
foreach my $option (@_)
{
my ($name, $value)=split(/=/, $option);
error("invalid sexpr option '$name'") if $name!~/^[dfhs]$|^raw$|^rate$/;
$sexprDebug=$value if $name eq 'd';
$sexprFilename=$value if $name eq 'f';
$sexprDispatch=1 if $name eq 'raw';
$sexprDispatch=2 if $name eq 'rate';
$sexprSubsys=$value if $name eq 's';
help() if $name eq 'h';
}
error("sexpr must be called with at least 'raw' or 'rate'") if !$sexprDispatch;
error("sexpr does not support -sJ") if $subsys=~/J/;
error("sexpr,raw does not apply to -sf") if $subsys=~/f/ && $sexprDispatch==1;
# the naming/location of the output files is based on code in lexpr
# If importing data, and if not reporting anything else, $subsys will be ''
$sexprSumFlag=$sexprSubsys=~/[cdfilmnstxE]/ ? 1 : 0;
error("sexpr subsys options '$sexprSubsys' not a proper subset of '$subsys'")
if $subsys ne '' && $sexprSubsys!~/^[$subsys]+$/;
error("sexpr cannot write a snapshot file and use a socket at the same time")
if $sockFlag && $sexprFilename ne '';
# Using -f and f= will not result in raw or plot file so need this message.
error ("using sexpr option 'f=' AND -f requires -P and/or --rawtoo")
if $sexprFilename ne '' && $filename ne '' && !$plotFlag && !$rawtooFlag;
# if -f, use that dirname/L for snampshot file; otherwise use f= for it.
$sexprFilename=(-d $filename) ? "$filename/S" : dirname($filename)."/S"
if $sexprFilename eq '' && $filename ne '';
print "sexpr filename: $sexprFilename\n" if $sexprDebug & 1;
if ($subsys=~/j/ && $subsys!~/C/)
{
logmsg('W',"adding -s+C because 'sexpr' requires it with 'j'");
$subsys.='C';
$CFlag=1;
}
}
sub sexprRaw
{
sexprHeaderPrint() if !$sexprHeaderPrinted;
# 1 extra level of indent (looks prettier) for XC
my $pad=$XCFlag ? ' ' : '';
my $sumFlag=$subsys=~/[cdfilmnstx]/ ? 1 : 0;
my $detFlag=$subsys=~/[CDN]/ ? 1 : 0;
my $cpuSumString=$cpuDetString='';
if ($subsys=~/c/i)
{
if ($subsys=~/c/)
{
my ($uTot, $nTot, $sTot, $iTot, $wTot, $irTot, $soTot, $stTot, $intTot)=(0,0,0,0,0,0,0,0);
for (my $i=0; $i<$NumCpus; $i++)
{
$uTot+= $userLast[$i];
$nTot+= $niceLast[$i];
$sTot+= $sysLast[$i];
$iTot+= $idleLast[$i];
$wTot+= $waitLast[$i];
$irTot+= $irqLast[$i];
$soTot+= $softLast[$i];
$stTot+= $stealLast[$i];
}
$cpuSumString.="$pad(cputotals (user $uTot) (nice $nTot) (sys $sTot) (idle $iTot) (wait $wTot) ";
$cpuSumString.= "(irq $irTot) (soft $soTot) (steal $stTot))\n";
$cpuSumString.="$pad(ctxint (ctx $ctxtLast) (int $intrptLast) (proc $procLast) (runq $loadQue))\n";
}
if ($subsys=~/C/)
{
my ($name, $userTot, $niceTot, $sysTot, $idleTot, $waitTot, $irqTot, $softTot, $stealTot, $intTot)=
('','','','','','','','','','');
for (my $i=0; $i<$NumCpus; $i++)
{
$name.= "cpu$i ";
$userTot.= "$userLast[$i] ";
$niceTot.= "$userLast[$i] ";
$sysTot.= "$userLast[$i] ";
$idleTot.= "$userLast[$i] ";
$waitTot.= "$userLast[$i] ";
$irqTot.= "$irqLast[$i] ";
$softTot.= "$softLast[$i] ";
$stealTot.="$stealLast[$i] ";
$intTot.= "0 "; # raw has no meaning since INTs totalled over all line entries
}
$name=~s/ $//; $userTot=~s/ $//; $niceTot=~s/ $//;
$sysTot=~s/ $//; $idleTot=~s/ $//; $waitTot=~s/ $//;
$irqTot=~s/ $//; $softTot=~s/ $//; $stealTot=~s/ $//;
$irqTot=~s/ $//;
$cpuDetString.="$pad(cpuinfo\n";
$cpuDetString.="$pad (name $name)\n";
$cpuDetString.="$pad (user $userTot)\n";
$cpuDetString.="$pad (nice $niceTot)\n";
$cpuDetString.="$pad (sys $sysTot)\n";
$cpuDetString.="$pad (idle $idleTot)\n";
$cpuDetString.="$pad (wait $waitTot))\n";
$cpuDetString.="$pad (irq $irqTot))\n";
$cpuDetString.="$pad (soft $softTot))\n";
$cpuDetString.="$pad (steal $stealTot))\n";
$cpuDetString.="$pad (int $intTot))\n";
}
}
my $diskSumString=$diskDetString='';
if ($subsys=~/d/i)
{
if ($subsys=~/d/)
{
my ($dRTot, $dRkbTot, $dWTot, $dWkbTot)=(0,0,0,0);
for (my $i=0; $i<@dskOrder; $i++)
{
# preserve display order but skip any disks not seen this interval
$dskName=$dskOrder[$i];
next if !defined($dskSeen[$i]);
next if ($dskFiltKeep eq '' && $dskName=~/$dskFiltIgnore/) || ($dskFiltKeep ne '' && $dskName!~/$dskFiltKeep/);
$dRTot+= $dskFieldsLast[$i][0];
$dRkbTot+= $dskFieldsLast[$i][2];
$dWTot+= $dskFieldsLast[$i][4];
$dWkbTot+= $dskFieldsLast[$i][6];
}
$diskSumString.="$pad(disktotals (reads $dRTot) (readkbs $dRkbTot) (writes $dWTot) (writekbs $dWkbTot))\n";
}
if ($subsys=~/D/)
{
my ($dName, $dRTot, $dRkbTot, $dWTot, $dWkbTot)=('','','','','');
for (my $i=0; $i<@dskOrder; $i++)
{
# preserve display order but skip any disks not seen this interval
$dskName=$dskOrder[$i];
next if !defined($dskSeen[$i]);
next if ($dskFiltKeep eq '' && $dskName=~/$dskFiltIgnore/) || ($dskFiltKeep ne '' && $dskName!~/$dskFiltKeep/);
$dName.= "$dskName ";
$dRTot.= "$dskFieldsLast[$i][0] ";
$dRkbTot.= "$dskFieldsLast[$i][2] ";
$dWTot.= "$dskFieldsLast[$i][4] ";
$dWkbTot.= "$dskFieldsLast[$i][6] ";
}
$dName=~s/ $//;
$dRTot=~s/ $//; $dRkbTot=~s/ $//;
$dWTot=~s/ $//; $dWkbTot=~s/ $//;
$diskDetString.="$pad(diskinfo\n";
$diskDetString.="$pad (name $dName)\n";
$diskDetString.="$pad (reads $dRTot)\n";
$diskDetString.="$pad (readkbs $dRkbTot)\n";
$diskDetString.="$pad (writes $dWTot)\n";
$diskDetString.="$pad (writekbs $dWkbTot))\n";
}
}
my $nfsString='';
if ($subsys=~/f/)
{
}
my $inodeString='';
if ($subsys=~/i/)
{
$inodeString= "$pad(inodeinfo (dentrynum $dentryNum) (dentryunused $dentryUnused) ";
$inodeString.="(inodeinfo.filesalloc $filesAlloc) (inodeinfo.filesmax $filesMax) (inodeinfo.inodeused $inodeUsed))\n";
}
# No lustre details, at least not for now...
my $lusSumString='';
if ($subsys=~/l/)
{
if ($CltFlag)
{
my ($reads, $readKBs, $writes, $writeKBs)=(0,0,0,0);
for (my $i=0; $i<$NumLustreFS; $i++)
{
$reads+= $lustreCltReadLast[$i];
$readKBs+= $lustreCltReadKBLast[$i];
$writes+= $lustreCltWriteLast[$i];
$writeKBs+=$lustreCltWriteKBLast[$i];
}
$lusSumString.="$pad(lusclt (reads $reads) (readkbs $readKBs) (writes $writes) (writekbs $writeKBs) (numfs $NumLustreFS))\n";
}
if ($OstFlag)
{
my ($reads, $readKBs, $writes, $writeKBs)=(0,0,0,0);
for (my $i=0; $i<$NumOst; $i++)
{
$reads+= $lustreReadOpsLast[$i];
$readKBs+= $lustreReadKBytesLast[$i];
$writes+= $lustreWriteOpsLast[$i];
$writeKBs+=$lustreWriteKBytesLast[$i];
}
$lusSumString.="$pad(lusoss (reads $reads) (readkbs $readKBs) (writes $writes) (writekbs $writeKBs))\n";
}
if ($MdsFlag)
{
my $getattrPlus=$lustreMdsGetattrLast+$lustreMdsGetattrLockLast+$lustreMdsGetxattrLast;
my $setattrPlus=$lustreMdsReintSetattrLast+$lustreMdsSetxattrLast;
my $varName=($cfsVersion lt '1.6.5') ? 'reint' : 'unlink';
my $varVal= ($cfsVersion lt '1.6.5') ? $lustreMdsReintLast : $lustreMdsReintUnlinkLast;
$lusSumString.="$pad(lusmds (getattrP $getattrPlus) (setattrP $setattrPlus) ";
$lusSumString.="(sync $lustreMdsSyncLast) ($varName $varVal))\n";
}
}
my ($memString, $memDetString)=('','');
if ($subsys=~/m/)
{
$memString= "$pad(meminfo (memtot $memTot) (memused $memUsed) (memfree $memFree) ";
$memString.="(memshared $memShared) (membuf $memBuf) (memcached $memCached) ";
$memString.="(memslab $memSlab) (memmap $memMap) (meminact $memInact) ";
$memString.="(memhugetot $memHugeTot) (memhugefree $memHugeFree) (memhugersvd $memHugeRsvd) ";
$memString.="(swaptot $swapTotal) (swapused $swapUsed) (swapfree $swapFree))\n";
}
if ($subsys=~/M/)
{
my ($names,$used,$free,$slab,$map,$anon,$lock,$act,$inact)=('','','','','','','','','');
for (my $i=0; $i<$CpuNodes; $i++)
{
$names.=" $i";
$free.= " $numaMem[$i]->{free}";
$used.= " $numaMem[$i]->{used}";
$slab.= " $numaMem[$i]->{slab}";
$map.= " $numaMem[$i]->{map}";
$anon.= " $numaMem[$i]->{anon}";
$lock.= " $numaMem[$i]->{lock}";
$act.= " $numaMem[$i]->{act}";
$inact.= " $numaMem[$i]->{inact}";
}
$memDetString= "$pad(numainfo\n";
$memDetString.=" (name$names)\n";
$memDetString.=" (free$free)\n";
$memDetString.=" (used$used)\n";
$memDetString.=" (slab$slab)\n";
$memDetString.=" (map$map)\n";
$memDetString.=" (anon$anon)\n";
$memDetString.=" (anon$lock)\n";
$memDetString.=" (inact$inact))\n";
}
my $netSumString=$netDetString='';
if ($subsys=~/n/i)
{
if ($subsys=~/n/)
{
my ($kbinT, $pktinT, $kboutT, $pktoutT)=(0,0,0,0);
for ($i=0; $i<@netOrder; $i++)
{
$netName=$netOrder[$i];
next if !defined($netSeen[$i]);
next if ($netFiltKeep eq '' && $netName=~/$netFiltIgnore/) || ($netFiltKeep ne '' && $netName!~/$netFiltKeep/);
next if $netName=~/lo|sit|bond/;
$kbinT+= $netRxKBLast[$i];
$pktinT+= $netRxPktLast[$i];
$kboutT+= $netTxKBLast[$i];
$pktoutT+=$netTxPktLast[$i];
}
$netSumString= "$pad(nettotals (netkbin $kbinT) (netpktin $pktinT) (netkbout $kboutT) (netpktout $pktoutT))\n";
}
if ($subsys=~/N/)
{
my ($name, $kbinT, $pktinT, $kboutT, $pktoutT)=('','','','','');
for ($i=0; $i<@netOrder; $i++)
{
$netName=$netOrder[$i];
next if !defined($netSeen[$i]);
next if ($netFiltKeep eq '' && $netName=~/$netFiltIgnore/) || ($netFiltKeep ne '' && $netName!~/$netFiltKeep/);
next if $netName=~/lo|sit/;
$name.= "$netName ";
$kbinT.= "$netRxKBLast[$i] ";
$pktinT.= "$netRxPktLast[$i] ";
$kboutT.= "$netTxKBLast[$i] ";
$pktoutT.="$netTxPktLast[$i] ";
}
$name=~s/ $|://g; $kbinT=~s/ $//; $pktinT=~s/ $//;
$kboutT=~s/ $//; $pktoutT=~s/ $//;
$netDetString= "$pad(netinfo\n";
$netDetString.="$pad (name $name)\n";
$netDetString.="$pad (netkbin $kbinT)\n";
$netDetString.="$pad (netpktin $pktinT)\n";
$netDetString.="$pad (netkbout $kboutT)\n";
$netDetString.="$pad (netpktout $pktoutT))\n";
}
}
my $sockString='';
if ($subsys=~/s/)
{
$sockString= "$pad(sockinfo (sockused $sockUsed) (socktcp $sockTcp) (sockorphan $sockOrphan) (socktw $sockTw) (sockalloc $sockAlloc) (sockmem $sockMem)";
$sockString.="(sockudp $sockUdp) (sockraw $sockRaw) (sockfrag $sockFrag) (sockfragm $sockFragM))\n";
}
my $tcpString='';
if ($subsys=~/t/)
{
$tcpString="$pad(tcpinfo ";
$tcpString.=" (iperrs $ipErrors)" if $tcpFilt=~/i/;
$tcpString.=" (tcperrs $tcpErrors)" if $tcpFilt=~/t/;
$tcpString.=" (udperrs $udpErrors)" if $tcpFilt=~/u/;
$tcpString.=" (icmperrs $icmpErrors)" if $tcpFilt=~/c/;
$tcpString.=" (tcpxerrs $tcpExErrors)" if $tcpFilt=~/T/;
$tcpString.=")\n";
}
my $intString='';
if ($subsys=~/x/i)
{
my ($kbInT, $pktInT, $kbOutT, $pktOutT)=(0,0,0,0);
for (my $i=0; $i<$NumXRails; $i++)
{
$kbInT+= $elanRxMBLast[$i]*1024;
$pktInT+= $elanRxLast[$i];
$kbOutT+= $elanTxMBLast[$i]*1024;
$pktOutT+=$elanTxLast[$i];
}
$port=$HCAPortActive;
for (my $i=0; $i<$NumHCAs; $i++)
{
$kbInT+= $ibFieldsLast[$i][$port][13];
$pktInT+= $ibFieldsLast[$i][$port][15];
$kbOutT+= $ibFieldsLast[$i][$port][12];
$pktOutT+=$ibFieldsLast[$i][$port][14];
}
$intString="$pad(iconnect (intkbin $kbInT) (intpktin $pktInT) (intkbout $kbOutT) (intpktout $pktOutT))\n";
}
my $envDetString='';
if ($subsys=~/E/i)
{
foreach $key (sort keys %$ipmiData)
{
$envDetString.="(env$key";
for (my $i=0; $i{$key}}); $i++)
{
my $name=$ipmiData->{$key}->[$i]->{name};
my $inst=($key!~/power/ && $ipmiData->{$key}->[$i]->{inst} ne '-1') ? $ipmiData->{$key}->[$i]->{inst} : '';
$envDetString.=" ($name$inst $ipmiData->{$key}->[$i]->{value})";
}
$envDetString.=")\n";
}
}
# if doing --import, those modules may wish to supply some s-expr output
my ($impSumString, $impDetString)=('','');
for (my $i=0; $i<$impNumMods; $i++) { &{$impPrintExport[$i]}('s', \$impSumString, \$impDetString, 'raw'); }
$sumFlag=1 if $impSumString ne '';
$detFlag=1 if $impDetString ne '';
# Build up as a single string
$sexprRec='';
$sexprRec.="(collectl_summary\n" if $XCFlag && $sumFlag;
$sexprRec.="$pad(sample (time $lastSecs[$rawPFlag]))\n" if $sumFlag;
$sexprRec.="$cpuSumString$diskSumString$nfsString$inodeString$memString$netSumString";
$sexprRec.="$lusSumString$sockString$tcpString$intString$impSumString";
$sexprRec.=")\n" if $XCFlag && $sumFlag;
$sexprRec.="$pad(sample (time $lastSecs[$rawPFlag]))\n" if !$sumFlag;
$sexprRec.="(collectl_detail\n" if $XCFlag && $detFlag;
$sexprRec.="$cpuDetString$diskDetString$memDetString$netDetString$envDetString$impDetString";
$sexprRec.=")\n" if $XCFlag && $detFlag;
# Either send data over socket or print to terminal OR write to
# a file, but not both!
if ($sockFlag || $sexprFilename eq '')
{
printText($sexprRec, 1); # include EOL marker at end
}
elsif ($sexprFilename ne '')
{
open SEXPR, ">$sexprFilename" or logmsg("F", "Couldn't create '$sexprFilename'");
print SEXPR $sexprRec;
close SEXPR;
}
}
sub sexprRate
{
sexprHeaderPrint() if !$sexprHeaderPrinted;
# 1 extra level of indent (looks prettier) for XC
my $pad=$XCFlag ? ' ' : '';
my $sumFlag=$subsys=~/[cdfilmnstx]/ ? 1 : 0;
my $detFlag=$subsys=~/[CDN]/ ? 1 : 0;
my $cpuSumString=$cpuDetString='';
if ($subsys=~/c/i)
{
if ($subsys=~/c/)
{
# CPU utilization is a % and we don't want to report fractions
my $i=$NumCpus;
$cpuSumString.=sprintf("$pad(cputotals (user %d) (nice %d) (sys %d) (wait %d) (irq %d) (soft %d) (steal %d) (idle %d))\n",
$userP[$i], $niceP[$i], $sysP[$i], $waitP[$i], $irqP[$i], $softP[$i], $stealP[$i], $idleP[$i]);
$cpuSumString.=sprintf("$pad(ctxint (ctx %d) (int %d) (proc %d) (runq $loadQue))\n",
$ctxt/$intSecs, $intrpt/$intSecs, $proc/$intSecs);
}
if ($subsys=~/C/)
{
my ($name, $userTot, $niceTot, $sysTot, $waitTot, $irqTot, $softTot, $stealTot, $idleTot, $intTot)=
('','','','','','','','','','');
for (my $i=0; $i<$NumCpus; $i++)
{
$name.= "cpu$i ";
$userTot.= "$userP[$i] ";
$niceTot.= "$niceP[$i] ";
$sysTot.= "$sysP[$i] ";
$waitTot.= "$waitP[$i] ";
$irqTot.= "$irqP[$i] ";
$softTot.= "$softP[$i] ";
$stealTot.="$stealP[$i] ";
$idleTot.= "$idleP[$i] ";
$intTot.= "$intrptTot[$i] ";
}
$name=~s/ $//; $userTot=~s/ $//; $niceTot=~s/ $//;
$sysTot=~s/ $//; $waitTot=~s/ $//; $irqTot=~s/ $//;
$softTot=~s/ $//; $stealTot=~s/ $//; $idleTot=~s/ $//;
$intTot=~s/ $//;
$cpuDetString.="$pad(cpuinfo\n";
$cpuDetString.="$pad (name $name)\n";
$cpuDetString.="$pad (user $userTot)\n";
$cpuDetString.="$pad (nice $niceTot)\n";
$cpuDetString.="$pad (sys $sysTot)\n";
$cpuDetString.="$pad (wait $waitTot))\n";
$cpuDetString.="$pad (irq $irqTot)\n";
$cpuDetString.="$pad (soft $softTot)\n";
$cpuDetString.="$pad (steal $stealTot)\n";
$cpuDetString.="$pad (idle $idleTot)\n";
$cpuDetString.="$pad (int $intTot)\n";
}
}
my $diskSumString=$diskDetString='';
if ($subsys=~/d/i)
{
if ($subsys=~/d/)
{
$diskSumString.=sprintf("$pad(disktotals (reads %d) (readkbs %d) (writes %d) (writekbs %d))\n",
$dskReadTot/$intSecs, $dskReadKBTot/$intSecs,
$dskWriteTot/$intSecs, $dskWriteKBTot/$intSecs);
}
if ($subsys=~/D/)
{
my ($dName, $dRTot, $dRkbTot, $dWTot, $dWkbTot)=('','','','','');
for (my $i=0; $i<@dskOrder; $i++)
{
# preserve display order but skip any disks not seen this interval
$dskName=$dskOrder[$i];
next if !defined($dskSeen[$i]);
next if ($dskFiltKeep eq '' && $dskName=~/$dskFiltIgnore/) || ($dskFiltKeep ne '' && $dskName!~/$dskFiltKeep/);
$dName.= "$dskName ";
$dRTot.= sprintf("%d ", $dskRead[$i]/$intSecs);
$dRkbTot.= sprintf("%d ", $dskReadKB[$i]/$intSecs);
$dWTot.= sprintf("%d ", $dskWrite[$i]/$intSecs);
$dWkbTot.= sprintf("%d ", $dskWriteKB[$i]/$intSecs);
}
$dName=~s/ $//;
$dRTot=~s/ $//; $dRkbTot=~s/ $//;
$dWTot=~s/ $//; $dWkbTot=~s/ $//;
$diskDetString.="$pad(diskinfo\n";
$diskDetString.="$pad (name $dName)\n";
$diskDetString.="$pad (reads $dRTot)\n";
$diskDetString.="$pad (readkbs $dRkbTot)\n";
$diskDetString.="$pad (writes $dWTot)\n";
$diskDetString.="$pad (writekbs $dWkbTot))\n";
}
}
my $nfsString='';
if ($subsys=~/f/)
{
$nfsString=sprintf("$pad(nfsinfo (reads %d) (writes %d) (meta %d) (commits %d)\n",
$nfsReadsTot/$intSecs, $nfsWritesTot/$intSecs, $nfsCommitTot/$intSecs, $nfsMetaTot/$intSecs);
}
my $inodeString='';
if ($subsys=~/i/)
{
$inodeString= "$pad(inodeinfo (dentrynum $dentryNum) (dentryunused $dentryUnused) ";
$inodeString.="(inodeinfo.filesalloc $filesAlloc) (inodeinfo.filesmax $filesMax) (inodeinfo.inodeused $inodeUsed))\n";
}
# No lustre details, at least not for now...
my $lusSumString='';
if ($subsys=~/l/)
{
if ($CltFlag)
{
$lusSumString.=sprintf("$pad(lusclt (reads %d) (readkbs %d) (writes %d) (writekbs %d) (numfs $NumLustreFS))\n",
$lustreCltReadTot/$intSecs, $lustreCltReadKBTot/$intSecs,
$lustreCltWriteTot/$intSecs, $lustreCltWriteKBTot/$intSecs);
}
if ($OstFlag)
{
$lusSumString.=sprintf("$pad(lusoss (reads %d) (readkbs %d) (writes %d) (writekbs %d))\n",
$lustreReadOpsTot/$intSecs, $lustreReadKBytesTot/$intSecs,
$lustreWriteOpsTot/$intSecs, $lustreWriteKBytesTot/$intSecs);
}
if ($MdsFlag)
{
my $getattrPlus=$lustreMdsGetattr+$lustreMdsGetattrLock+$lustreMdsGetxattr;
my $setattrPlus=$lustreMdsReintSetattr+$lustreMdsSetxattr;
my $varName=($cfsVersion lt '1.6.5') ? 'reint' : 'unlink';
my $varVal= ($cfsVersion lt '1.6.5') ? $lustreMdsReint : $lustreMdsReintUnlink;
$lusSumString.=sprintf("$pad(lusmds (getattrP %d) (setattr %d) (sync %d) ($varName %d)\n",
$getattrPlus/$intSecs, $setattrPlus/$intSecs,
$lustreMdsSync/$intSecs, $varVal/$intSecs);
}
}
my ($memString,$memDetString)=('','');
if ($subsys=~/m/)
{
$memString= "$pad(meminfo (memtot $memTot) (memused $memUsed) (memfree $memFree) ";
$memString.="(memshared $memShared) (membuf $memBuf) (memcached $memCached) ";
$memString.="(memslab $memSlab) (memmap $memMap) (meminact $memInact) ";
$memString.="(memhugetot $memHugeTot) (memhugefree $memHugeFree) (memhugersvd $memHugeRsvd) ";
$memString.="(swaptot $swapTotal) (swapused $swapUsed) (swapfree $swapFree)\n";
}
if ($subsys=~/M/)
{
my ($names,$used,$free,$slab,$map,$anon,$act,$inact)=('','','','','','','','');
for (my $i=0; $i<$CpuNodes; $i++)
{
$names.=" $i";
$free.= " $numaMem[$i]->{free}";
$used.= " $numaMem[$i]->{used}";
$slab.= " $numaMem[$i]->{slab}";
$map.= " $numaMem[$i]->{map}";
$anon.= " $numaMem[$i]->{anon}";
$act.= " $numaMem[$i]->{act}";
$inact.= " $numaMem[$i]->{inact}";
}
$memDetString= "$pad(numainfo\n";
$memDetString.=" (name$names)\n";
$memDetString.=" (free$free)\n";
$memDetString.=" (used$used)\n";
$memDetString.=" (slab$slab)\n";
$memDetString.=" (map$map)\n";
$memDetString.=" (anon$anon)\n";
$memDetString.=" (inact$inact))\n";
}
my $netSumString=$netDetString='';
if ($subsys=~/n/i)
{
if ($subsys=~/n/)
{
$netSumString=sprintf("$pad(nettotals (netkbin %d) (netpktin %d) (netkbout %d) (netpktout %d))\n",
$netRxKBTot/$intSecs, $netRxPktTot/$intSecs,
$netTxKBTot/$intSecs, $netTxPktTot/$intSecs);
}
if ($subsys=~/N/)
{
my ($name, $kbinT, $pktinT, $kboutT, $pktoutT)=('', '','','','');
for ($i=0; $i<@netOrder; $i++)
{
$netName=$netOrder[$i];
next if !defined($netSeen[$i]);
next if ($netFiltKeep eq '' && $netName=~/$netFiltIgnore/) || ($netFiltKeep ne '' && $netName!~/$netFiltKeep/);
next if $netName=~/lo|sit/;
$name.= "$netName ";
$kbinT.= sprintf("%d ", $netRxKB[$i]/$intSecs);
$pktinT.= sprintf("%d ", $netRxPkt[$i]/$intSecs);
$kboutT.= sprintf("%d ", $netTxKB[$i]/$intSecs);
$pktoutT.=sprintf("%d ", $netTxPkt[$i]/$intSecs);
}
$name=~s/ $|://g; $kbinT=~s/ $//; $pktinT=~s/ $//;
$kboutT=~s/ $//; $pktoutT=~s/ $//;
$netDetString= "$pad(netinfo\n";
$netDetString.="$pad (name $name)\n";
$netDetString.="$pad (netkbin $kbinT)\n";
$netDetString.="$pad (netpktin $pktinT)\n";
$netDetString.="$pad (netkbout $kboutT)\n";
$netDetString.="$pad (netpktout $pktoutT))\n";
}
}
my $sockString='';
if ($subsys=~/s/)
{
$sockString= "$pad(sockinfo (sockused $sockUsed) (socktcp $sockTcp) (sockorphan $sockOrphan) (socktw $sockTw) (sockalloc $sockAlloc) (sockmem $sockMem)";
$sockString.="(sockudp $sockUdp) (sockraw $sockRaw) (sockfrag $sockFrag) (sockfragm $sockFragM))\n";
}
my $tcpString='';
if ($subsys=~/t/)
{
$tcpString="$pad(tcpinfo ";
$tcpString.=" (iperrs $ipErrors)" if $tcpFilt=~/i/;
$tcpString.=" (tcperrs $tcpErrors)" if $tcpFilt=~/t/;
$tcpString.=" (udperrs $udpErrors)" if $tcpFilt=~/u/;
$tcpString.=" (icmperrs $icmpErrors)" if $tcpFilt=~/c/;
$tcpString.=" (tcpxerrs $tcpExErrors)" if $tcpFilt=~/T/;
$tcpString.=")\n";
}
my $intString='';
if ($subsys=~/x/i)
{
if ($NumXRails)
{
$kbInT= $elanRxKBTot;
$pktInT= $elanRxTot;
$kbOutT= $elanTxKBTot;
$pktOutT=$elanTxTot;
}
if ($NumHCAs)
{
$kbInT= $ibRxKBTot;
$pktInT= $ibRxTot;
$kbOutT= $ibTxKBTot;
$pktOutT=$ibTxTot;
}
$intString=sprintf("$pad(iconnect (intkbin %d) (intpktin %d) (intkbout %d) (intpktout %d))\n",
$kbInT/$intSecs, $pktInT/$intSecs, $kbOutT/$intSecs, $pktOutT/$intSecs);
}
my $envDetString='';
if ($subsys=~/E/i)
{
foreach $key (sort keys %$ipmiData)
{
$envDetString.="(env$key";
for (my $i=0; $i{$key}}); $i++)
{
my $name=$ipmiData->{$key}->[$i]->{name};
my $inst=($key!~/power/ && $ipmiData->{$key}->[$i]->{inst} ne '-1') ? $ipmiData->{$key}->[$i]->{inst} : '';
$envDetString.=" ($name$inst $ipmiData->{$key}->[$i]->{value})";
}
$envDetString.=")\n";
}
}
# if doing --import, those modules may wish to supply some s-expr output
my ($impSumString, $impDetString)=('','');
for (my $i=0; $i<$impNumMods; $i++) { &{$impPrintExport[$i]}('s', \$impSumString, \$impDetString, 'rate'); }
$sumFlag=1 if $impSumString ne '';
$detFlag=1 if $impDetString ne '';
$sexprRec='';
$sexprRec.="(collectl_summary\n" if $XCFlag && $sumFlag;
$sexprRec.="$pad(sample (time $lastSecs[$rawPFlag]))\n" if $sumFlag;
$sexprRec.="$cpuSumString$diskSumString$nfsString$inodeString$memString$netSumString";
$sexprRec.="$lusSumString$sockString$tcpString$intString$impSumString";
$sexprRec.=")\n" if $XCFlag && $sumFlag;
$sexprRec.="(collectl_detail\n" if $XCFlag && $detFlag;
$sexprRec.="$pad(sample (time $lastSecs[$rawPFlag]))\n" if !$sumFlag;
$sexprRec.="$cpuDetString$diskDetString$memDetString$netDetString$envDetString$impDetString";
$sexprRec.=")\n" if $XCFlag && $detFlag;
# Either send data over socket or print to terminal OR write to
# a file, but not both!
if ($sockFlag || $sexprFilename eq '')
{
printText($sexprRec, 1);
}
elsif ($sexprFilename ne '')
{
open SEXPR, ">$sexprFilename" or logmsg("F", "Couldn't create '$sexprFilename'");
print SEXPR $sexprRec;
close SEXPR;
}
}
sub sexprHeaderPrint
{
# 1 extra level of indent (looks prettier) for XC
my $pad=$XCFlag ? ' ' : '';
my $sumFlag=$subsys=~/[cdfilmnstx]/ ? 1 : 0;
my $detFlag=$subsys=~/[CDN]/ ? 1 : 0;
$sexprHdr='';
$sexprHdr.="(collect_summary\n" if $XCFlag && $sumFlag;
$sexprHdr.="$pad(sample (time var))\n";
$sexprHdr.="$pad(cputotals (user val) (nice val) (sys val) (idle val) (wait val) (irq val) (soft val) (steal val))\n"
if $subsys=~/c/;
$sexprHdr.="$pad(ctxint (ctx val) (int val) (proc val) (runq val))\n"
if $subsys=~/c/;
$sexprHdr.="$pad(disktotals (reads val) (readkbs val) (writes val) (writekbs val))\n"
if $subsys=~/d/;
$sexprHdr.="$pad(nfsinfo (read val) (write val) (calls val))\n"
if $subsys=~/f/;
$sexprHdr.="$pad(inodeinfo (unuseddcache val) (openfiles val) (inodeused val) (superuer val)(dquotused val))\n"
if $subsys=~/i/;
$sexprHdr.="$pad(lusclt (reads val) (readkbs val) (writes val) (writekbs val) (numfs val))\n"
if $subsys=~/l/ && $CltFlag;
$sexprHdr.="$pad(lusmds (close val) (getattr val) (reint val) (sync val))\n"
if $subsys=~/l/ && $MdsFlag;
$sexprHdr.="$pad(lusoss (reads val) (readkbs val) (writes val) (writekbs val))\n"
if $subsys=~/l/ && $OstFlag;
$sexprHdr.="$pad(meminfo (memtot val) (memused val) (memfree val) (memshared val) (membuf val) (memcached val) (memslab val) (memmap val) (meminact val))\n"
if $subsys=~/m/;
$sexprHdr.="$pad(nettotals (netkbin val) (netpktin val) (netkbout val) (netpktout val))\n"
if $subsys=~/n/;
$sexprHdr.="$pad(sockinfo (sockused val) (socktcp val) (sockorphan val) (socktw val) (sockalloc val) (sockmem val)(sockudp val) (sockraw val) (sockfrag val) (sockfragm val))\n"
if $subsys=~/s/;
if ($subsys=~/t/)
{
$sexprHdr.="$pad(tcpinfo";
$sexprHdr.=" (iperrs val)" if $tcpFilt=~/i/;
$sexprHdr.=" (tcperrs val)" if $tcpFilt=~/t/;
$sexprHdr.=" (udperrs val)" if $tcpFilt=~/u/;
$sexprHdr.=" (icmperrs val)" if $tcpFilt=~/c/;
$sexprHdr.=" (tcpxerrs val)" if $tcpFilt=~/T/;
$sexprHdr.=")\n";
}
$sexprHdr.="$pad(iconnect (intkbin val) (intpktin val) (intkbout val) (intpktout val))\n"
if $subsys=~/x/;
$sexprHdr.=")\n" if $XCFlag && $sumFlag;
$sexprHdr.="(collect_detail \n" if $XCFlag && $subsys=~/[CDN]/;
if ($subsys=~/C/)
{
my $names='';
$sexprHdr.="$pad(cpuinfo\n";
for (my $i=0; $i<$NumCpus; $i++)
{
$names.="cpu$i ";
}
$sexprHdr.="$pad (name $names)\n";
$sexprHdr.="$pad (user $names)\n";
$sexprHdr.="$pad (nice $names)\n";
$sexprHdr.="$pad (sys $names)\n";
$sexprHdr.="$pad (idle $names)\n";
$sexprHdr.="$pad (wait $names)\n";
$sexprHdr.="$pad (irq $names)\n";
$sexprHdr.="$pad (soft $names)\n";
$sexprHdr.="$pad (steal $names)\n";
$sexprHdr.="$pad (int $names)\n";
}
if ($subsys=~/D/)
{
my $names='';
$sexprHdr.="$pad(diskinfo\n";
for (my $i=0; $i<@dskOrder; $i++)
{
# preserve display order but skip any disks not seen this interval
$dskName=$dskOrder[$i];
next if !defined($dskSeen[$i]);
next if ($dskFiltKeep eq '' && $dskName=~/$dskFiltIgnore/) || ($dskFiltKeep ne '' && $dskName!~/$dskFiltKeep/);
$names.="$dskName ";
}
$sexprHdr.="$pad (name $names)\n";
$sexprHdr.="$pad (reads $names)\n";
$sexprHdr.="$pad (readkbs $names)\n";
$sexprHdr.="$pad (writes $names)\n";
$sexprHdr.="$pad (writekbs $names)\n";
}
if ($subsys=~/N/)
{
my $names='';
$sexprHdr.="$pad(netinfo\n";
for ($i=0; $i<@netOrder; $i++)
{
$netName=$netOrder[$i];
next if !defined($netSeen[$i]);
next if ($netFiltKeep eq '' && $netName=~/$netFiltIgnore/) || ($netFiltKeep ne '' && $netName!~/$netFiltKeep/);
next if $netName=~/lo|sit/;
$names.="$netName ";
}
$names=~s/://g;
$sexprHdr.="$pad (name $names)\n";
$sexprHdr.="$pad (netkbin $names)\n";
$sexprHdr.="$pad (netpktin $names)\n";
$sexprHdr.="$pad (netkbout $names)\n";
$sexprHdr.="$pad (netpktout $names)\n";
}
$sexprHdr.=")\n" if $detFlag && $XCFlag;
# The header only goes to a file
if ($sexprFilename ne '')
{
my $headerFilename=sprintf("%s/#", dirname($sexprFilename));
print "Create sexpr header file: $headerFilename\n" if $sexprDebug & 2;
open SEXPR, ">$headerFilename" or logmsg("F", "Couldn't create '$headerFilename/#'");
print SEXPR $sexprHdr;
close SEXPR;
$sexprHeaderPrinted=1;
}
}
sub help
{
my $text=< $address,
PeerPort => $port,
Proto => 'tcp',
Timeout =>1);
if (!defined($socket))
{
print "Couldn't connect to server, retrying...\n";
sleep 1;
next;
}
print "Socket opened on $address:$port\n";
$select = new IO::Select($socket);
while ($socket ne '')
{
$buffer='';
while (my @ready=$select->can_read(10))
{
$bytes=sysread($socket, $line, 100);
#print "BYTES: $bytes\n";
if ($bytes==0)
{
print "Socket closed on other end\n";
$socket='';
last;
}
$buffer.=$line;
@handles=($select->can_read(0));
last if scalar(@handles)==0;
}
print "$buffer" if $buffer ne '';
}
}
sub sigInt
{
print "Close Socket\n";
$socket->close() if defined($socket) && $socket ne '';
exit;
}
collectl-3.6.9/man1/ 0000775 0001750 0001750 00000000000 12230241726 012333 5 ustar mjs mjs collectl-3.6.9/man1/collectl.1 0000644 0001750 0001750 00000124735 12230241727 014231 0 ustar mjs mjs .TH COLLECTL 1 "APRIL 2003" LOCAL "Collectl" -*- nroff -*-
.SH NAME
collectl - Collects data that describes the current system status.
.SH SYNOPSIS
Record Mode - read data from live system and write to file or display on terminal
.B collectl [\-f file] [options]
Playback Mode \- read data from one or more raw data files and display
on terminal
.B collectl \-p file1 [file2 ...] [options]
.SH OPTIONS
Record Mode
In this mode data is taken from a
.BR live
system and either displayed on the
terminal or written to one or more files or a socket.
.B "--align"
.RS
If the HiRes modules is present,
.BR collectl
sample monitoring will be aligned such that a sample will always be taken at the
top of a minute (this does NOT mean the first sample will occur then) so that all
instances of collectl running on any systems which have their clocks synchronized
will all take samples at the same time. Furthermore, if one is doing process
monitoring, those samples will also be taken at the top of the minute and so can
delay the start of sampling up to 2 full process monitoring intervals.
.RE
.B "--all"
.RS
Collect summary data for ALL subsystems except slabs, since slab monitoring requires
a different monitoring interval. This also means you won't get any detail data which
also includes processes and environmementals.
You can use this switch anywhere \-s can be used
but not both together. If the system supports lustre and/or interconnect monitoring
those statistics will be provided but the warnings produced when they are not
available you try to select them with \-s will not be displayed.
.RE
.B "\-A, --address address[:port[:timeout]] | server[:port]"
.RS
In the first form, one specifies an address, optional port
and timeout (the first colon is required to specify timeout for default port).
All data is then written to that socket prefaced with the current host
name at the named address and port until
the socket is closed, at which time collectl will exit.
In the second form one enters the text "server" and optional port.
In this form, collectl runs as a server, waiting for a connection and
once established writes data on that socket. The key difference here is
if the client exists collectl keeps running and will again look for a
new connection, allowing it to survive client restarts or crashes.
The default port is set at 2655 but can be changed \- see collectl.conf.
In both forms, one can additionally request local data logging by
specifying a combination of \-P and \-f. See
.B "man collectl-logging"
for more details.
.RE
.B "\--comment string"
.RS
Add the specified string to the end of the headers in the data files. If any embedded
spaces be sure to quote it. This can be very useful when doing characterizations or
benchmarking and you're frequently changing system/application parameters and restarting
collectl between tests.
.RE
.B "\-C, --config filename"
.RS
Name/location of the collectl configuration file.
If not specified,
.BR collectl
searches for
.BR collectl.conf
first in /etc (the default), then in the same directory the
.BR collectl
executable is in, and finally the current working directory.
.RE
.B "\-c, --count Samples"
.RS
The number of samples to record. This is one way of 3 ways of describing
how long collectl should run (see
.BR \-r
and
.BR \-R
). Note that these 3 switches are mutually exclusive.
.RE
.B "\-D, --daemon"
.RS
Run
.BR collectl
as a daemon, primarily used when starting as a service. One
caveat about this mode is you can only run one copy.
.RE
.B "--export file[,options]"
.RS
This requests that collectl does not print anything on the terminal (or
send it to a socket) using the standard brief/verbose/plot formats.
Instead it executes a perl "require" on the named file, using an extension of
ph if not specified. It first looks in the current directory and if not
there the directory the executable is in. It then calls the
function "file"Init(options) towards the beginning of collectl and again as
simply "file"(@options) to generate the exported formatted output. See the
online documentation on Exporting Custom Output and Logging for more details.
.RE
.B "\-f, --filename Filename"
.RS
This is the name of a file to write the output to. See the
description of
.BR File
.BR Naming
for further details.
.RE
.B \-F, --flush seconds
.RS
Flush output buffers after this number of seconds. This is equivalent to
issuing
.B kill \-s USR1
at the same frequency (but a lot easier!). If 0, a flush will occur every
data collection interval.
.RE
.B --grep pattern
.RS
The main purpose of this switch is for those users who have discovered there is
some data in the raw files that never appears in any display and have taken to
displaying it themselves with grep. Unfortunately this method does not include
timestamps and so makes it difficult to interpret the results. Even if you
include the timestamp from the file it is in UTC and so needs to be translated
to be of any real value. This switch does just that and then some.
Specifically, it allows you to playback a file and instead of processing it normally
it simply searches for any entries that match the perl pattern and reports those
lines prefaced with time stamps. You can optionally change the time format
with the usual \-o options and can even select the timeframe with --from and --thru.
.RE
.B --home
.RS
Always start the display for the current interval at the top of the screen
also known as the home position (non-plot format only). This generates a
real-time, continously refreshing display when the data fits on a single screen.
.RE
.B --import file1[,options][:file2[,options]...]
.RS
This loads the named files and executes callbacks to them, which is the API mechanism for
importing additional metrics into collectl. See the webpage on the API for further detail.
Since these files also include instructions for how to report the output in all the various
forms, you will also need to include --import during playback. Finally, since the
default is to seamlessly include imported data with everything else collectl reports, if you ONLY
want to display imported data you much explicitly deselect all other subsystems either
by including -s- (note the trailing minus sign) followed by all the subsystems were recorded OR
simply say -s-all.
.RE
.B "\-i, --interval interval[:interval2[:interval3]]"
.RS
This is the sampling interval in seconds. The default is 10 seconds when run
as a daemon and 1 second otherwise. The process subsystem and slabs (\-sY and \-sZ)
are sampled at the lower rate of
.BR interval2.
Environmentals (\-sE), which only apply to a subset of hardware, are sampled at
.BR interval3.
Both
.BR interval2
and
.BR interval3,
if specified, must be an even multiple of
.BR interval1.
The daemon default is \-i10:60:300 and all other modes are \-i1:60:300.
To sample only processes once every 10 seconds
use \-i:10.
.RE
.B --nohup
.RS
Whenever collectl finishes a data collection interval, it checks to see if the starting parent
has exited. This is to prevent the case in which someone might start a copy of collectl
and then the process dies and collectl keeps running. If that is the behavior someone
actually intends, they should start collectl with --nohup.
NOTE - when running as a daemon, --nohup is implied.
.RE
.B "--quiet"
.RS
Whenever collectl wants to tell the user something, it assigns a category to it such as
Informational, Warning, Error or Fatal. When run with \-m, all messages are displayed
for the user and if logging data to a file with \-f, these messages are also sent to a
log file which is in the data collection directory and has an extenion of "log".
However, if \-m is not specified Informational messages (such as collectl starting
or stopping) are not reported on the terminal but the other 3 are. Sometimes the
warnings can be annoying and one can suppress these with --quiet though they will still be
written to the message log in \-f. You cannot suppress Error or Fatal errors.
.RE
.B "\-r, --rolllogs time[[,days[:months]][,minutes]]"
.RS
When selected, collectl
runs indefinately (or at least until the system reboots).
The maximum number of raw and/or plot files that will be retained
(older ones are automatically deleted) is controlled by the
.BR days
field, the default is 7. When -m is also specified to direct collectl to
write messages to a log file in the logging directory, the number of months
to retain those logs is controlled by the
.BR months
field and its default is 12. The
.BR increment
field which is also optional (but is position dependent) specifies the duration of
an individual collection file in minutes the default of which is 1440 or 1 day.
.RE
.B "--rawdskfilt"
.RS
This switch overrides the DiskFilter setting in collectl.conf and explicitly defines
a perl regx expression against which records from /prod/diskstats are selected for
processing. When there are a lot of disks to process, this can be a handy way to reduce
the amount of data collected and actually improve performance since there are less
patterns to match each input record against.
Just remember that unlike --dskfilt which only filters during display, records filtered
with this switch are never even recorded and so lost forever.
As a side benefit of this switch, if you really want to look at partition level stats
you can do so by leaving off the trailing space in the default pattern.
One must be also be careful in selecting the correct pattern since it's easy to get it wrong
and you may end up collecting the WRONG data! To verify you are collecting what you think
you are, make a test run using -d4 to see the raw data being recorded in real-time.
.RE
.B "--rawdskignore"
.RS
This is the opposite of the rawdskfilt switch. When specified any disks listed are completely
ignored and will not appear in the raw file. Typically this switch is useful when you're only
interested in recording a subset of disk statistics.
.RE
.B "--rawnetfilt"
.RS
This works just like --rawdskfilt except it applies to networks. Unlike disk filtering
which has an explicit default pattern, the default for network filtering is to simply
record all network data from /proc/net/dev.
The -d4 switch also works here, as well as everywhere, to see the raw data as it is being
collected.
.RE
.B "--rawnetignore"
.RS
This is the opposite of the rawnetfilt switch and works just like the rawdskignore switch.
When specified any networks listed are ignored and will not appear in the raw file.
Typically this switch is useful when you're only interested in recording a subset of network
statistics.
.RE
.B "--rawtoo"
.RS
Only available in conjunction with \-P, this switch causes the creation/logging
of raw data in addition to plottable data. While this may seem excessive,
keep in mind that unlike plottable data, raw data can be played back with different
switches potentially providing more details. The overhead to write out this
additional data is minimal, the only real cost being that of extra disk space.
.RE
.B "\-R, --runas uid[:gid]"
.RS
This switch only works when running in daemon mode and so must be specified in
the DaemonCommands line. Its presence will cause collectl to write the collectl.pid
file into the same directory as its other output files as specified by -f, since
/var/run does not normally grant non-privileged users write access. Furthermore,
the ownership of that directory must match the specified ownership since collectl
needs to write ALL it's files to that directory and can no longer assume global
permissions when run as root.
This WILL also require manually modifying /etc/init.d/collectl to change the
PIDFILE variable to point to the same directory which the -f switch in the
DaemonCommands line of collectl.conf points to.
As a final note of caution, since this mechanism changes where collectl reads/writes
its pid file, once you start using --runas, all calls to run collectl as a daemon
must use it or it may be confused and exhibit unpredictable behavior.
.RE
.B "\-R, --runtime duration"
.RS
Specify the duration of data collection where the duration is a number followed
by one of
.BR wdhms,
indicating how many weeks, days, hours, minutes or seconds
the collection is to be taken for.
.RE
.B "--sep separator"
.RS
Specify the plot format separator \- default is a space. If this is a numeric field it is
interpretted as the decimal value of the associated ASCII character code. Otherwise it
is interpretted as the character itself. In other words, "--sep :" sets the separator
character to a colon and "--sep 9" sets it to a horizontal tab. "--sep 58" would also
set it to a colon.
.RE
.B --tworaw
.RS
The switches \-G and --group have been replaced by --rawtoo, which is more rescriptive of
its function. When specified, it tells collectl to treat process and slab data as an
entirely separate group of raw files, named with the extention "rawp". These separate files
can be played back and processed just like any other collectl raw files and in fact one
can even play back both at the same time if that is what is
desired. The only real purpose of this switch is that on some systems with many processes, it is
possible to generate huge raw files (some have been observerd to be >250MB!) and while collectl
will happily play back/process these files it can take a long time. By using the --tworaw switch one
still gets a huge rawp file, but the normal raw file is a much more manageable size and as
a result will faster to process then when all data is combined into the same file.
.RE
Playback Mode
In this mode, data is read from one or more data files that were
generated in Record Mode
.B "--export Filename"
.RS
When playing back a file, use this switch to create an identical raw file differing only in
the timeframe being convered, so naturally one must also include --from, --thru or both.
Further, since the resultant file will contain the exact same raw data you cannot select a
subset using \-s. This switch is actually intended for a support function for situations
where somone is having problems playing back a file and a subset of the original raw file
that covers the problem time has been requested, hopefully allowing a significantly
file to be posted or emailed.
.RE
.B "--extract filename"
.RS
If specified, rather than actually play back the file specified with \-p, ALL raw data between
the date ranges is selected and a subset of that raw file created. The rules for how to interpret
the filename are the same as used for \-f.
.RE
.B "\-f, --filename filename"
.RS
If specified, this is the name of a file or directory
to write the output to (rather than
the terminal). See the description for details on the format of this field.
This requires the \-P flag as well.
.RE
.B "--from time range"
.RS
Play back data starting with this time, which may optionally include the ending
time as well, which is of the format of [date:]time[-[date:]time].
The leading 0 of the hour is optional and if the seconds field is not specified
is assumed to be 0. If no dates specified the time(s) apply to each file specified
by \-P. Otherwise the time(s) only apply to the first/last dates and any files
between those dates will have all their data reported.
.RE
.B "--offsettime seconds"
.RS
This field originally was used before collectl reported the timezone in the file
headers and allowed one to compensate. Since then it is rarely needed except in
two possible cases, one in which data on two systems is to be compared and they
weren't synchonized with ntp. This allows all the times to be reported as shifted
by some number of seconds. The other case (and this is very rare) is when a clock
had changed in the middle of a sample and will not be converted correctly. When
this happens one may have to play back the samples in pieces and manually set
the time offset.
.RE
.B "--passwd filename"
.RS
When reporting usernames associated with a UID, use this file for the mapping.
This is particularly important on systems running NIS where this are no user
names in /etc/passwd.
.RE
.B "\-p, --playback Filename"
.RS
Read data from the specified
.BR playback
file(s), noting that one can use wildcards in the filename if
quoted (if playing back multiple files to the terminal you probably
want to include \-m to see the filenames as they are processed).
The filename must either end
in
.BR raw
or
.BR raw.gz.
As an added feature, since people sometimes automate
the running of this option and don't want to hard code a date, you can
specify the string YESTERDAY or TODAY and they will be replaced in the
filename string by the appropriate date.
.RE
.RE
.B "--pname name"
.RS
By default, collectl uses the file /var/run/collectl.pid to indicate the pid of
the running instance of collectl and prevent multiple copies from being run. If
you DO want to run a second copy, this switch will cause collectl to change its
process name to collectl-name and use that name as the associated pid file as well.
.RE
.B --procanalyze
.RS
When specified and there is process data in the raw file, a summary file will be generated
with one entry unique process containing such things as the total cpu consumed for both user
and system, min/max utilization of various memory types, total page faults and several others.
.RE
.B --slabanalyze
.RS
When specified and there is slab data in the raw file, a summary file will be generated
with one entry unique slab containing data on physical memory usage by that slab.
.RE
.B "--thru time"
.RS
Time thru which to play back a raw file. See --from for more
.RE
Common Switches \- both record and playback modes
.RE
.B "\-d, --debug debug"
.RS
Control the level of debugging information, not typically used. For details
see the source code.
.RE
.B \-h, --help, \-x, --helpext, \-X, --helpall
.RS
Display standard, extended help message (which doesn't include the optional displays
such as --showoptions, --showsubsys, --showsubopts, --showtopopts) or everything.
.RE
.B --hr, --headerrepeat num
.RS
Sets the number of intervals to display data for before repeating the header.
A value \-1 will prevent any headers from being displayed and a value of 0
will cause only a single header to be displayed and never repeated.
.RE
.B --iosize
.RS
In brief mode, include iosize with disk, infiniband and network data.
.RE
.B \-l, --limits limit
.RS
Override one or more default exception limits. If more than one limit they
must be separated by hyphens. Current values are:
.B SVC:value
.RS
Report partition activity with Service times >= 30 msec
.RE
.B IOS:value
.RS
Report device activity with 10 or more reads or writes per second
.RE
.B LusKBS:value
.RS
Report client or OSS activity greater than limit. Only applies to
Client Summary or OSS Detail reporting. [default=100000]
.RE
.B LusReints:value
.RS
Report MDS activity with Reint greater than limit. Only applies
to MDS Summary reporting. [default=1000]
.RE
.B AND
.RS
Both the IOS and SCV limits must be reached before a device is reported. This
is the default value and is only included for completeness.
.RE
.B OR
.RS
Report device activity if either IOS or SVC thresholds are reached.
.RE
.B \-L, --lustsvcs [c|m|o][:seconds]
.RS
This switch limits which servics lustre checks for and the frequency of those checks.
For more information see the man page collectl\-lustre.
.RE
.RE
.B \-m, --messages
.RS
Write status to a monthly log file in the same directory as the output file
(requires \-f to be specified as well). The name of the file will be
.BR collectl\-yyyymm.log
and will track various messages that may get generated during every run of
.BR collectl.
.RE
.B \-N, --nice
.RS
Set priority to a
.BR nicer
one of 10.
.RE
.B "\-o, --options Options"
.RS
These apply to the way output is displayed OR written to a plot file. They
do not effect the way data is selected for recording. Most of these switches
work in both record as well as playback mode. If you're not sure, just
try it.
.B 1
.RS
Data in plotting format should use 1 decimal point of precision as appropriate.
.RE
.B 2
.RS
Data in plotting format should use 2 decimal points of precision as appropriate.
.RE
.B a
.RS
Always append data to an existing plot file. By default if a plot
file exists, the playback file will be skipped as a way of assuring it is
associated with a single recorded file. This switch overrides that mechanism
allowing muliple recorded files to be processed and written to a single plot
file.
.RE
.B c
.RS
Always open newly named plot fies in
.BR create
mode, overwriting any old ones
that may already exists. If one processes multiple files for the same day in
.BR append
mode multiple times, the same data will be appended to the same file mulitple
times. This assures a new file is created at the start of the processing.
.RE
.B d
.RS
For use with terminal output and brief mode. Preceed each line with a date/time stamp,
the date being in mm/dd format. This option can also be applied to plot formatit
which will cause the date portion to also be displayed in this format as
opposed to D format.
.RE
.B D
.RS
For use with terminal output and brief mode. Preceed each line with a date/time
stamp, the date being in yyyymmdd format.
.RE
.B g
.RS
For use with terminal output and brief mode. When displaying values of 1G or greater
there is limited precision for 1 digit values. This options provides a way to display
additional digits for more granularity by substituting a "g" for the decimal point
rather than the trailing "G".
.RE
.B G
.RS
For use with terminal output and brief mode. This is similar to "g" but preserves
the trailing "G" by sacrificing a digit of granularity.
.RE
.B m
.RS
Whenever times are reported in plot format, in the normal
terminal reporting format at the bginning of each interval or when when one
of the time reporting options (d, D, T or U is selected), append the milliseconds
to the time.
.RE
.B n
.RS
Where appropriate, data such as disk KBs or transfers are normalized to units per
second by taking the change in a counter and dividing by the number of seconds in
that interval. In the case of CPUs, utilization (calculated in jiffies) is
normalized as a percentage of the interval.
Normalization can be disabled via this option, the result being
the reported values are not divided by the duration of the interval. This can be
particulary useful for reporting values that are < 1/2 the sampling, which will be
rounded to 0.
.RE
.B T
.RS
For use with terminal output and brief mode, preceeds each line with a time stamp.
.RE
.B u
.RS
Create plot files with unique names by include the starting time of a colletion
in the name. This forces
multiple collections taken the same day to be written to multiple files.
.RE
.B "\-U or --utc"
.RS
In plot format only, report timestamps in Coordinated Universal time which is more
commonly know as UTC.
.RE
.B x
.RS
Report only exception records for selected subsystems. Exception reporting also requires
--verbose. Currently this only
applies to disk detail and Lustre server information so one must select at least
-s D, l or L for
this to apply. If writing to a detail file, this data will go into a separate
file with the extension
.BR X
appended to the regular detail file name.
.RE
.B X
.RS
Report both exceptions as well as all details for selected subsystems, for
-s D, l or L only.
.RE
.B z
.RS
If the compression library has been installed, all output files will be compressed by
default. This switch tells collectl not to compress any plottable files. If collectl
tries to compress but cannot because the library hasn't been installed, it will generate
a warning which can be suppressed with this switch.
.RE
.RE
.RE
.B \-P, --plot
.RS
Generate output in plot format. This format is space separated data which
consists of a header (prefaced with a # for easy identification by an analysis
program as well as identifying it as a comment for programs, such as gnuplot,
which honor that convention). When written to disk, which is the typical way
this option is used,
.BR summary
data elements
are written to the
.BR tab
file and the
.BR detail
elements written to one or
more files, one per detail subsystem.
If \-f is not specified, all output is sent to the terminal.
Output is always one line per sampling interval.
.RE
.B "--stats"
.RS
This switch will cause brief data to be reported as both totals and averages
after processing one or more files for the same day or in playback mode.
.RE
.B "--statopts option(s)"
.RS
This switch controls the way brief stats are reported, the default is to report the totals once, at the end of a
day's worth of raw files, if more than one.
.br
a \- include averages along with totals
.br
i \- include the interval data itself, which is the equivalent of -oA
.br
s \- print summary stats at the end of each file processed even if more than one per day
.RE
.B "\-s, --subsys subsystem"
.RS
This field controls which subsystem data is to be collected or played back
for. The rules for displaying results vary depending on the type of data to be
displayed. If you write data for CPUs and DISKs to a raw file and play it back
with \-sc, you will only see CPU data. If you play it back with \-scm you will
still only see CPU data since memory data was not collected. However, when
used with \-P, collectl will always honor the subsystems specified with
this switch so in the previous example you will see CPU
data plus memory data of all 0s. To see the current set of default subsystems,
which are a subset of this full list,
use \-h.
You can also use + or \- to add or subtract subsystems to/from the default values.
For example, "\-s\-cdn+N"< will remove cpu, disk and network monitoring from the
defaults while adding network detail.
The default is "cdn", which stands for CPU, Disk and Network summary data.
Refer to data definitions on the sourceforge website OR in
/usr/share/collectl/doc/collectl\-xxx to see complete descriptions of the data returned.
SUMMARY SUBSYSTEMS
.br
b \- buddy info (memory fragmentation)
.br
c \- CPU
.br
d \- Disk
.br
f \- NFS V3 Data
.br
i \- Inode and File System
.br
j \- Interrupts
.br
l \- Lustre
.br
m \- Memory
.br
n \- Networks
.br
s \- Sockets
.br
t \- TCP
.br
x \- Interconnect
.br
y \- Slabs (system object caches)
DETAIL SUBSYSTEMS
This is the set of
.BR detail
data from which in most cases the corresponding summary data is
derived. There are currently 2 types that do not have corresponding summary
data and those are "Environmental" and "Process". So, if one has 3 disks
and chooses
.B \-sd,
one will only see a single total taken
across all 3 disks. If one
chooses
.B \-sD,
individual disk totals will be reported but no totals. Choosing
.B \-sdD
will get you both.
.br
C \- CPU
.br
D \- Disk
.br
E \- Environmental data (fan, power, temp), via ipmitool
.br
F \- NFS Data
.br
J \- Interrupts
.br
L \- Lustre OST detail OR client Filesystem detail
.br
M \- Memory node data, which is also known as numa data
.br
N \- Networks
.br
T \- 65 TCP counters only available in plot format
.br
X \- Interconnect
.br
Y \- Slabs (system object caches)
.br
Z \- Processes
.RE
.B --showheader
.RS
In collectl mode this command will cause the header that is normally written to a data file to
be displayed on the terminal and collectl then exists. This can be a handy way to get a brief
overview of the system configuration.
.RE
.B --showoptions
.RS
This command shows only the portion of the help text that desribes the \-o and --options switches
to save the time of wading through the entire help screen.
.RE
.B --showcolheaders
.RS
This command shows the first set of headers that will be printed by collectl and exits. Doesn't
really make sense for multi-section output like several sets of verbose or detail data. Also
note that since it requires one monitoring interval to build up some headers which may be dynamic,
it also forces the interval to 0.
.RE
.B --showsubopts
.RS
List all the subsystem specifice options
.RE
.B --showtopopts
.RS
Show all the different values for the --top type field, which specify the
field(s) by to sort the data
.RE
.B --showrootslabs
.RS
This command only works on systems using the new slab allocator and will list the root
name (these are those entries in /sys/slab which are not soft links) along with all
its alias names. If a name doesn't have an alias, it will not appear in this report.
.RE
.B --showslabaliases
.RS
This command only works on systems using the new slab allocator. Like --showrootslabs, it
will name a slab and all its aliases but rather than show the root slab name
it will show one of the aliases to provide a more meaningful name. If there are any
slabs that only have a single (or no) alias they will not be included in this report.
.RE
.B --showsubopts
.RS
Similar to --showoptions, this command summaries just the paramaters associated with \-O and
--subopts.
.RE
.B --showsubsys
.RS
Yet another way to summare a portion of the help text, this command only shows valid subsystems.
.RE
.B "--top [type][,num]"
.RS
Include the top "num" consumers by resource for this interval. The default number is the height
of the window if it can be determined otherwise 24, and the
default resource is the total cpu time which is taken as the sum of SysT and UsrT.
See --showtopopts for a list of other types of data you can sort on.
This switch can also be used with \-s in which case a portion of the window is reserved at the
top to fill in the subsystem data, which is currently in verbose mode though a brief format
is contemplated for some time in the future.
In interactive mode and if not specified, the process
monitoring interval will be set to that for other subsystems. The screen will be cleared for each interval
resulting in a display similar to the "top" utility. In playback more the screen will NOT be cleared. You
cannot use this switch in "record" mode.
.RE
.B "--umask mask"
.RS
Sets collectl's umask to control output file permissions. Only root can set
the umask. See "man umask" for details.
.RE
.B "--utime mask"
.RS
Write periodic micro-timestamps into raw file at different points in time for
fine grained measurements of operation times.
.br
1 \- write timestamps when entering major sections
.br
2 \- write timestamps for all /proc accesses except for process data
.br
4 \- write timestamps for /proc data for all processes including threads
.RE
.B \-v
.RS
Show version and whether or not Compression and/or HiResTime modules have
been installed and exit.
.RE
.B \-V
.RS
Show default parmeter and control settings, all of which can be changed in
/etc/collectl.conf
.RE
.B --verbose
.RS
Display output in verbose mode. This often displays more data than in the default mode. When
displaying detail data, verbose mode is forced. Furthermore, if summary data for a single
subsystem is to be displayed in verbose mode, the headers are only repeated occasionally whereas
if multiple subsystems are involved each needs their own header.
.RE
.B \-w
.RS
Disply data in
.BR wide
mode. When displaying data on the terminal, some data is formatted followed
by a K, M or G as appropriate. Selecting this switch will cause the
full field to be displayed. Note that there is no attempt
to align data with the column headings in this mode.
.RE
.SH SUBSYSTEM OPTIONS
The following options are subsystem specific and typically filter data for collection
and/or display as well as affect the output format:
.B "--cpuopts"
.RS
.br
z \- only applies to cpu details, do not report any CPUs with no load. In other
words all entries are zero except for IDLE.
.RE
.B "--dskfilt [^]perl-regx[,perl-regx...]"
.RS
NOTE \- this does NOT effect data collection, ALL disk data will always be collected.
However, only data for disk names that match the pattern(s) will be included in the summary
totals and displayed when details are requested. Alternatively, if you preface the
first expression with a caret, all names that match all strings will be excluded from the summary
totals and detail displays rather then included. If you don't know perl, a partial string will
usually work too.
.RE
.B "--dskopts"
.RS
.br
f \- report some columns as fractions for more precision on detail output
.br
i \- display the i/o sizes in brief mode just like with --iosize
.br
o \- exclude unused disks from new file headers and plot data
.br
z \- only applies to disk details, do not report any lines with values of all zeros.
.RE
.B "--envopts Environmental Options"
.RS
The default is to display ALL data but the following will cause a subset to be displayed
.br
f \- display fan data
.br
p \- display current (power) data
.br
t \- display temperature data
.br
C \- convert temperature to Celcius if in Farenheit
.br
F \- convert temperature to Farenheit if in Celcius
.br
M \- display each type of data on separate line
.br
T \- display data truncated to whole integers (some implemenations
displayed them with fractional components)
.br
9 \- any number, will tell ipmitool to read on this device number
.RE
.B "--envfilt regx"
If specified, this regx is evaluated against each line of data returned by ipmitool
and only those that match are retained. All other data is lost.
.RS
.RE
.B "--envremap perl-regx,..."
.RS
If specified as a comma separated list of perl regular substitution expressions without the
=~s portion, each expression is applied to each environmental field name, thereby allowing
one to rename the column headers. This can be most useful when running on heterogeneuos
systems and you want consistent column names.
.RE
.B "--intfilt [^]perl-regx[,perl-regx...]"
.RS
NOTE \- this does NOT effect data collection, ALL interrupt data will always be collected.
However, only data for interrupts that match the pattern(s) will be included in the summary
totals and displayed when details are requested. Alternatively, if you preface the
first expression with a caret, all names that match all strings will be excluded from the summary
totals and detail displays rather then included. If you don't know perl, a partial string will
usually work too.
NOTE - these expressions are applied to the entire line one sees in /proc/interrupts, including
the interrupt number, name and even counters so if you do want to include an interrupt number
in the pattern be sure to include the trailing colon as well.
.RE
.B "--lustopts Lustre Options"
.RS
.br
B \- For clients and servers, show buffer stats
.br
D \- For MDSs and OSTs AND running earlier versions of HPSFS, collect disk block iostats
.br
M \- For clients, collect metadata
.br
O \- For OSTs, show detail level stats
.br
R \- For client, collect readahead stats
.RE
.B "--memopts Memory Options"
.RS
R \- show memory values (including swap space) as rates of change as opposed to absolute
values. One can also show absolute changes between intervals by including \-on.
.RE
.B "--netfilt [^]perl-regx[,perl-regx...]"
.RS
NOTE \- this does NOT effect data collection, ALL network data will always be collected.
However, only data for network names that match the pattern(s) will be included in the summary
totals and displayed when details are requested. Alternatively, if you preface the
first expression with a caret, all names that match all strings will be excluded from the summary
totals and detail displays rather then included. If you don't know perl, a partial string will
usually work too.
.RE
.B "--netopts"
.RS
e \- include network error counts in brief and explicit error types elsewhere
.br
E \- only include lines with network errors in them
.br
i \- include i/o sizes in brief mode
.br
o \- exclude unused networks from new file headers and plot data
.br
w \- set width of network device name
.RE
.B "--nfsfilt NFS Filters"
.RS
Specify one or more comma separated filters as a C/S followed by an nfs
version number and only those will have data reported on. For example,
C2 says to report data on V2 Clients. As a data collection performance
optimization, if one or more client filters are specified, data will
actually be collected for all clients as is also done for servers.
.RE
.B "--nfsopts NFS Options"
q.RS
z \- only display detail lines which have data
.RE
.B "--procfilt Process Filters"
.RS
These filters restrict which processes are selected for collection/display.
Using this filter will significanly reduce the load on process data collection
since collectl creates a blacklist of those existing processes that do not
pass the filter and so are permanently excluded from any future processing.
The format of a filter is a one charter type followed by a match string.
Multiple filters may be specified if separated by commas.
.br
c \- substring of the command being executed as explicitly read from /proc/pid/stat.
Note that this can actually be a perl expression, so if you want a command that ends
in a particular string all you need to is append a \$ to the end of the string.
Otherwise it would match any commands containing that string.
.br
C \- any command that starts with the specified string
.br
f \- full path of the command, including arguments, as read from /proc/pid/cmdline. Like
the c modifier this too can be a perl expression.
.br
p \- pid
.br
P \- parent pid
.br
u \- any process ownerd by this user's UID or in the range specifide by uxxx\-yyy
.br
U \- any process owned by this username
.B "caution:"
the process names collectl tries to match with c and C is the second field in /proc/pid/stat
which may not necessarily be what you think! eg the name for X emacs is actually emacs-x
.RE
.B --procopts options
.RS
These options control the way data is displayed and can also improve data
collection performance
.br
c \- include CPU time of children who have exited (same as ps \-S)
.br
f \- use cumulative totals for page faults in process data instead of rates
.br
i \- show process I/O counters in display instead of default format
.br
I \- disable collection of I/O counters, see note below
.br
k \- remove known shells from process names, making it possible to see actual command
.br
m \- show breakdown of memory utilization instead of default format
.br
p \- never look for new pids or threads during data collection
.br
r \- show root command name only (no directory) for narrower display
.br
R \- show ALL process priorities ('RT' currently displayed if realtime)
.br
s \- show process start time in hh:mm:ss format
.br
S \- show process start time in mmmdd-hh:mm:ss format
.br
t \- include ALL process threads (increases collection overhead)
.br
u \- report username as 12 chars instead of 8, noting uxx will cause column width
to be xx but cannot be less than 8
.br
w \- widen display by including whole argument string, with optional max width
.br
x \- include extended process attributes (currently only for context switches)
.br
z \- exclude any processes with 0 in sort field (in --top mode)
Process data is the most expensive type of data collected, costing as much as 3
times the CPU load as all other types of data combined. Collecting thread data
makes this even more expensive. One can significantly reduce this load by over
25 percent by disabling the collection of I/O stats. However, keep in mind that
even if you don't try to optimize process data collection, the overall system
load by collectl can still be on the order of about 0.2% when running as a daemon
with default collection rates. See the online documentation on measuring
performance for more information.
A security hole was identified that allowed non-priviledged users to read
/proc/pid/io and guess password lengths and noe many distros retrict
access to the owner or root. As a result, non-priviledged users will see
all 0 I/O counts for processes that are not theirs when specifying --procopt i.
.RE
.RE
.B "--slabfilt Slab Filters"
.RS
One can specify a list of slab names separated by commas and only those
slabs whose names start with those strings will be listed or summaried.
.RE
.B "--slabopts Slab Options"
.RS
.B "s \- exclude any slabs with an allocation of 0"
.RE
.RS
.B "S \- only show those slabs whose allocations changed since last display"
.RE
.B "--xopts"
.RS
i \- include i/o sizes in brief mode
.RE
.SH DESCRIPTION
The
.BR collectl
utility is a system monitoring tool that records or displays
specific operating system data for one or more sets of subsystems. Any set
of the subsystems, such as CPU, Disks, Memory or Sockets can
be included in or excluded from data collection. Data can either be
displayed back to the terminal, or stored in either a compressed or
uncompressed data file. The data files themselves can either be in
.BR raw
format
(essentially a direct copy from the associated /proc structures) or in a space
separated
.BR plottable
format such that it can be easily plotted using tools
such as gnuplot or excel. Data files can be read and manipulated from the
command line, or through use of command scripts.
Upon startup,
.BR collectl.conf
is read, which sets a number of default parameters and switch values. Collectl
searches for this file first in /etc, then in the directory the collectl
execuable lives in (typically /usr/sbin) and finally the current directory.
These locations can be overriden with the
.BR \-C
switch. Unless
you're doing something really special, this file need never be touched, the
only exception perhaps being when choosing to run collectl as a service and you
wish to change it's default behavior which is set by the DaemonCommand entry.
.SH RESTRICTIONS/PROBLEMS
Thread reporting currently only works with 2.6 kernels.
The pagesize has been hardcoded for perl 5.6 systems to 4096 for IA32
and 16384 for all others. If you are running 5.6 on a system with a
different pagesize you will see incorrect SLAB allocation sizes and will
need to scale the numbers you're seeing accordingly.
I have recently discovered there is a bug in /proc in that an extra line
is occasionally read with the end of the previous buffer! When this
occurs a message is written (if \-m enabled) and always written to the
terminal. Since this happens with a higher frequency with process data
I silently ignore those as the output can get pretty noisey.
If for any reason this is a problem, be sure to let me know.
Since collectl has no control over the frequency at which data gets written
to /proc, one can get anomolous statistics as collectl is only
reporting a snapshot of what is being recorded. For more information
see http://collectl.sourceforge.net/TheMath.html.
At least one network card occasionally generates erroneous network stats and to try
to keep the data rational, collectl tries to detect this and when it does generates
a message that bogus data has been detected.
.SH FILES, EXAMPLES AND MORE INFORMATION
http://collectl.sourceforge.net OR /opt/hp/collectl/docs
.SH ACKNOWLEDGEMENTS
I would like to thank Rob Urban for his creation of the Tru64 Unix
collect tool, which collectl is based on.
.SH AUTHOR
This program was written by Mark Seger (mjseger@gmail.com).
.br
Copyright 2003-2011 Hewlett-Packard Development Company, LP
.br
collectl may be copied only under the terms of either the Artistic License
or the GNU General Public License, which may be found in the source kit
collectl-3.6.9/vmstat.ph 0000644 0001750 0001750 00000003566 12230241726 013356 0 ustar mjs mjs sub vmstatInit
{
error("-s not allowed with 'vmstat'") if $userSubsys ne '';
error("-f requires either --rawtoo or -P") if $filename ne '' && !$rawtooFlag && !$plotFlag;
error("-P or --rawtoo require -f") if $filename eq '' && ($rawtooFlag || $plotFlag);
$subsys=$userSubsys='cm';
}
sub vmstat
{
my $line;
$sameColsFlag=1;
if (printHeader())
{
$line= "${clscr}#${miniBlanks}procs ---------------memory (KB)--------------- --swaps-- -----io---- --system-- ----cpu-----\n";
$line.="#$miniDateTime r b swpd free buff cache inact active si so bi bo in cs us sy id wa\n";
$headersPrinted=1;
}
$datetime='';
if ($options=~/[dDTm]/)
{
($ss, $mm, $hh, $mday, $mon, $year)=localtime($lastSecs[0]);
$datetime=sprintf("%02d:%02d:%02d", $hh, $mm, $ss);
$datetime=sprintf("%02d/%02d %s", $mon+1, $mday, $datetime) if $options=~/d/;
$datetime=sprintf("%04d%02d%02d %s", $year+1900, $mon+1, $mday, $datetime) if $options=~/D/;
$datetime.=".$usecs" if ($options=~/m/);
$datetime.=" ";
}
# currently only happens when called by colmux
if ($showColFlag)
{
printText($line);
return;
}
my $i=$NumCpus;
my $usr=$userP[$i]+$niceP[$i];
my $sys=$sysP[$i]+$irqP[$i]+$softP[$i]+$stealP[$i];
$line.=sprintf("%s %2d %2d %6s %6s %6s %6s %6s %6s %4d %4d %5d %5d %4d %5d %2d %2d %3d %2d\n",
$datetime, $procsRun, $procsBlock,
cvt($swapUsed,6,1,1), cvt($memFree,6,1,1), cvt($memBuf,6,1,1),
cvt($memCached,6,1,1), cvt($memInact,6,1,1), cvt($memAct,6,1,1),
$swapin/$intSecs, $swapout/$intSecs, $pagein/$intSecs, $pageout/$intSecs,
$intrpt/$intSecs, $ctxt/$intSecs,
$usr, $sys, $idleP[$i], $waitP[$i]);
printText($line);
}
1;
collectl-3.6.9/misc.ph 0000664 0001750 0001750 00000015643 12230241726 012774 0 ustar mjs mjs # copyright, 2003-2009 Hewlett-Packard Development Company, LP
# NOTE - by default, this module only collectl data once a minute, which you can change
# with the i=x parameter (eg --import misc,i=x). Regardless of the collection interval,
# date will be reported every interval in brief/verbose formats to provide a consisent
# set of output each monitoring cycle. However, --export lexpr will report light-weight
# counters every interval but heavy-weight ones (currenly only logins) based in i=.
# sexpr and gexpr will report all 4 counters every interval independent of when sampled.
# To report --export data in ALL lexpr samples, in case the listener expects it, include
# the a switch (eg misc,a).
# M i s c e l l a n e u o s C o u n t e r
use strict;
# Allow reference to collectl variables, but be CAREFUL as these should be treated as readonly
our ($miniFiller, $rate, $SEP, $datetime, $miniInstances, $interval, $showColFlag);
my (%miscNotOpened, $miscUptime, $miscMHz, $miscMounts, $miscLogins);
my ($miscUptimeTOT, $miscMHzTOT, $miscMountsTOT, $miscLoginsTOT);
my ($miscInterval, $miscImportCount, $miscSampleCounter, $miscAllFlag);
sub miscInit
{
my $impOptsref=shift;
my $impKeyref= shift;
# If we ever run with a ':' in the inteval, we need to be sure we're
# only looking at the main one. NOTE - if --showcolflag, collectl
# sets $interval to 0 and we need to make sure out division doesn't bomb
my $miscInterval1=(split(/:/, $interval))[0];
$miscInterval1=1 if $showColFlag;
# For now, only options are a, 'i=' and s
$miscInterval=60;
$miscAllFlag=0;
if (defined($$impOptsref))
{
foreach my $option (split(/,/,$$impOptsref))
{
my ($name, $value)=split(/=/, $option);
error("invalid misc option: '$name'") if $name ne 'a' && $name ne 'i' && $name ne 's';
$miscInterval=$value if $name eq 'i';
$miscAllFlag=1 if $name eq 'a';
}
}
$miscImportCount=int($miscInterval/$miscInterval1);
error("misc interval option not a multiple of '$miscInterval1' seconds")
if $miscInterval1*$miscImportCount != $miscInterval;
$$impOptsref='s'; # only one collectl cares about
$$impKeyref='misc';
$miscSampleCounter=-1;
$miscLogins=0;
return(1);
}
# Nothing to add to header
sub miscUpdateHeader
{
}
sub miscGetData
{
getProc(0, '/proc/uptime', 'misc-uptime');
grepData(1, '/proc/cpuinfo', 'MHz', 'misc-mhz');
grepData(2, '/proc/mounts', ' nfs ', 'misc-mounts');
# we only retrieve heavy-weight counters at the misc sampling interval
# as specified by "i=" or the default value of 60.
return if ($miscSampleCounter++ % $miscImportCount)!=0;
getExec(4, '/usr/bin/who -s -u', 'misc-logins');
}
sub miscInitInterval
{
}
sub miscAnalyze
{
my $type= shift;
my $dataref=shift;
$type=~/^misc-(.*)/;
$type=$1;
my @fields=split(/\s+/, $$dataref);
if ($type eq 'uptime')
{
$miscUptime=$fields[0];
}
elsif ($type eq 'mhz')
{
$miscMHz=$fields[3];
}
elsif ($type eq 'mounts')
{
$miscMounts=$fields[0];
}
elsif ($type eq 'logins:') # getExec adds on the ':'
{
$miscLogins=$fields[0];
}
}
sub miscPrintBrief
{
my $type=shift;
my $lineref=shift;
if ($type==1) # header line 1
{
$$lineref.="<------Misc------>";
}
elsif ($type==2) # header line 2
{
$$lineref.=" UTim MHz MT Log ";
}
elsif ($type==3) # data
{
$$lineref.=sprintf(" %4s %4d %2d %3d ",
cvt($miscUptime/86400), $miscMHz, $miscMounts, $miscLogins);
}
elsif ($type==4) # reset 'total' counters
{
$miscUptimeTOT=$miscMHzTOT=$miscMountsTOT=$miscLoginsTOT=0;
}
elsif ($type==5) # increment 'total' counters
{
$miscUptimeTOT+= int($miscUptime/86400+.5); # otherwise we get round off error
$miscMHzTOT+= $miscMHz;
$miscMountsTOT+= $miscMounts;
$miscLoginsTOT+= $miscLogins;
}
elsif ($type==6) # print 'total' counters
{
printf " %4d %4d %2d %3d ", $miscUptimeTOT/$miniInstances, $miscMHzTOT/$miniInstances,
$miscMountsTOT/$miniInstances, $miscLoginsTOT/$miniInstances;
}
}
sub miscPrintVerbose
{
my $printHeader=shift;
my $homeFlag= shift;
my $lineref= shift;
my $line='';
if ($printHeader)
{
$line.="\n" if !$homeFlag;
$line.="# MISC STATISTICS\n";
$line.="#$miniFiller UpTime CPU-MHz Mounts Logins\n";
}
$$lineref.=$line;
return if $showColFlag;
$$lineref.=sprintf("$datetime %6s %6d %6d %6d \n",
cvt($miscUptime/86400), $miscMHz, $miscMounts, $miscLogins);
}
sub miscPrintPlot
{
my $type= shift;
my $ref1= shift;
# Headers - note we end with $SEP but that's ok because writeData() removes it
$$ref1.="[MISC]Uptime${SEP}[MISC]MHz${SEP}[MISC]Mounts${SEP}[MISC]Logins${SEP}"
if $type==1;
# Summary Data Only - and here we start with $SEP
$$ref1.=sprintf("$SEP%d$SEP%d$SEP%d$SEP%d",
$miscUptime/86400, $miscMHz, $miscMounts, $miscLogins)
if $type==3;
}
sub miscPrintExport
{
my $type= shift;
my $ref1= shift;
my $ref2= shift;
my $ref3= shift;
# The light-weight counters are reported every sampling interval but since I think sexpr
# needs to be contant, we'll always report all even if some only sampled periodically.
# Same thing for gexpr, at least for now.
if ($type eq 'l')
{
push @$ref1, "misc.uptime"; push @$ref2, sprintf("%d", $miscUptime/86400);
push @$ref1, "misc.cpuMHz"; push @$ref2, sprintf("%d", $miscMHz);
push @$ref1, "misc.mounts"; push @$ref2, sprintf("%d", $miscMounts);
}
elsif ($type eq 's')
{
$$ref1.=sprintf(" (misctotals (uptime %d) (cpuMHz %d) (mounts %d) (logins %d))\n",
$miscUptime/86400, $miscMHz, $miscMounts, $miscLogins);
}
elsif ($type eq 'g')
{
push @$ref2, 'num', 'num', 'num', 'num', 'num';
push @$ref1, "misc.uptime"; push @$ref3, sprintf("%d", $miscUptime/86400);
push @$ref1, "misc.cpuMHz"; push @$ref3, sprintf("%d", $miscMHz);
push @$ref1, "misc.mounts"; push @$ref3, sprintf("%d", $miscMounts);
push @$ref1, "misc.logins"; push @$ref3, sprintf("%d", $miscLogins);
}
# Heavy-weight lexpr counters are only returned based on "i=" or default of 60 seconds
return if !$miscAllFlag && (($miscSampleCounter-1) % $miscImportCount)!=0;
if ($type eq 'l')
{
push @$ref1, "misc.logins"; push @$ref2, sprintf("%d", $miscLogins);
}
}
# Type 1: return contents of first match
# Type 2: return count of all matches
sub grepData
{
my $type= shift;
my $proc= shift;
my $string=shift;
my $tag= shift;
# From getProc()
if (!open PROC, "<$proc")
{
# but just report it once, but not foe nfs or proc data
logmsg("W", "Couldn't open '$proc'") if !defined($miscNotOpened{$proc});
$miscNotOpened{$proc}=1;
return(0);
}
my $count=0;
foreach my $line ()
{
next if $line!~/$string/;
if ($type==1)
{
record(2, "$tag $line");
return;
}
$count++;
}
record(2, "$tag $count\n");
}
1;
collectl-3.6.9/initd/ 0000775 0001750 0001750 00000000000 12230241727 012607 5 ustar mjs mjs collectl-3.6.9/initd/collectl-suse 0000755 0001750 0001750 00000007205 12230241727 015315 0 ustar mjs mjs #!/bin/sh
# Startup script for collectl on SuSE based distributions
#
# description: Run data collection for a number of subsystems
# see /etc/collectl.conf for startup options
# $2: process name, if other than collectl OR "" to use $3
# $3: parameters, in quotes if more than one, to pass to daemon command
#
### BEGIN INIT INFO
# Provides: collectl
# Required-Start: $ALL
# Required-Stop: $ALL
# Default-Start: 2 3 5
# Default-Stop: 0 1 6
# Short-Description: Collectl monitors system performance.
# Description: Collectl is a light-weight performance monitoring
# tool capable of reporting interactively as well
# as logging to disk. It reports statistics on
# cpu, disk, infiniband, lustre, memory, network,
# nfs, process, quadrics, slabs and more in easy
# to read format.
### END INIT INFO
# description: Run data collection for a number of subsystems
# see /etc/collectl.conf for startup options
# $2: process name extension, if other than collectl [optional for start/restart/reload]
# $3: parameters, in quotes if more than one, to pass to daemon command
#
# EXAMPLES:
# run at 1 second interval and only collect cpu/disk data
# /etc/init.d/collectl start "-i1 -scd"
# run a second instance with instance name of 'int5', with interval of 5 seconds
# /etc/init.d/collectl start int5 "-i5"
[ -r /etc/rc.status ] && . /etc/rc.status
rc_reset
COLLECTL=/usr/sbin/collectl
if [ ! -f $COLLECTL ]; then
echo -n "Cannot find $COLLECTL"
rc_status -s
rc_exit
fi
PATH=/bin:/usr/bin:/sbin:/usr/sbin
export PATH
# Just to make sure nothing is different when running 'collectl', we
# won't use --check even though it's probably ok to use all the time.
PNAME=collectl
if [ "$2" != "" ]; then
EXT=$2
if [ "$1" = "start" ] || [ "$1" = "restart" ] ; then
if [ "$3" = "" ]; then
SWITCHES=$2
EXT=""
else
SWITCHES=$3
fi
fi
# Just to make sure nothing is different when running 'collectl', we
# won't use --check even though it's probably ok to use all the time.
if [ "$EXT" != "" ]; then
PNAME="collectl-$EXT"
PSWITCH="--pname $EXT"
CHECK="--check $PNAME "
fi
fi
PROCNAME=$PNAME
PIDFILE="/var/run/$PNAME.pid"
# If a pidfile, make sure it's not stale and if it is, collectl not running
if [ -f $PIDFILE ]; then
pid=`cat $PIDFILE`
pid=`ps ax opid,cmd | grep $PROCNAME | grep $pid | grep -v grep | awk '{ print $1 }'`
fi
case "$1" in
start)
if [ "$pid" != "" ]; then
echo $PNAME already running
exit
fi
# we used to start with 'startproc', but if an instance of collectl already running it
# won't start the next one so we'll just start it this way
echo -n "Starting $PNAME:"
$COLLECTL -D $SWITCHES $PSWITCH
rc_status -v
;;
stop)
# Note that we need to use a pid file to identify which instance we want to stop
if [ -f $PIDFILE ]; then
echo -n "Shutting down $PNAME: "
killproc -p $PIDFILE collectl
rc_status -v
else
echo "$PNAME not running"
fi
;;
flush)
if [ -f $PIDFILE ]; then
pid=`cat $PIDFILE`
echo Flushing buffers for $PNAME
kill -s USR1 $pid
else
echo "$PNAME not running"
fi
rc_status
;;
status)
if [ "$pid" = "" ]; then
echo "$PNAME not running"
else
echo "$PNAME is running"
fi
;;
restart|reload)
$0 stop $EXT
$0 start "$2" "$3"
rc_status
;;
*)
echo "Usage: $0 {start|stop|flush|restart|status}"
exit 1
esac
rc_exit
collectl-3.6.9/initd/collectl 0000755 0001750 0001750 00000006317 12230241727 014343 0 ustar mjs mjs #!/bin/sh
# Startup script for collectl
#
### BEGIN INIT INFO
# Provides: collectl
# Required-Start: $all
# Required-Stop: $all
# Default-Start: 2 3 4 5
# Default-Stop: 0 1 6
# Short-Description: Collectl monitors system performance.
# Description: Collectl is a light-weight performance monitoring
# tool capable of reporting interactively as well
# as logging to disk. It reports statistics on
# cpu, disk, infiniband, lustre, memory, network,
# nfs, process, quadrics, slabs and more in easy
# to read format.
### END INIT INFO
# chkconfig: 345 99 99
# description: Run data collection for a number of subsystems
# see /etc/collectl.conf for startup options
# $2: process name extension, if other than collectl [optional for start/restart/reload]
# $3: parameters, in quotes if more than one, to pass to daemon command
#
# EXAMPLES:
# run at 1 second interval and only collect cpu/disk data
# /etc/init.d/collectl start "-i1 -scd"
# run a second instance with instance name of 'int5', with interval of 5 seconds
# /etc/init.d/collectl start int5 "-i5"
RETVAL=0
COLLECTL=/usr/bin/collectl
. /etc/rc.d/init.d/functions
# older distros do not support "status -p" and we need to know if this one does
STATUS=`status`
if [[ $STATUS = *-p* ]]; then IFLAG=1; else IFLAG=0; fi
if [ $IFLAG -eq 0 ] && [ "$3" != '' ]; then
echo "starting multiple instances with this distro is not supported";
exit
fi
if [ ! -f $COLLECTL ]; then
echo -n "Cannot find $COLLECTL"
exit 1
fi
PNAME=collectl
if [ "$2" != "" ]; then
EXT=$2
if [ "$1" = "start" ] || [ "$1" = "restart" ] || [ "$1" = "reload" ]; then
if [ "$3" = "" ]; then
SWITCHES=$2
EXT=""
else
SWITCHES=$3
fi
fi
# Just to make sure nothing is different when running 'collectl', we
# won't use --check even though it's probably ok to use all the time.
if [ "$EXT" != "" ]; then
PNAME="collectl-$EXT"
PSWITCH="--pname $EXT"
CHECK="--check $PNAME "
fi
fi
PIDFILE="/var/run/$PNAME.pid"
case "$1" in
start)
COMMAND="$COLLECTL -D $SWITCHES $PSWITCH"
echo -n "Starting $PNAME:"
daemon $CHECK$COMMAND
RETVAL=$?
echo
[ $RETVAL -eq 0 ] && touch /var/lock/subsys/$PNAME
;;
stop)
if [ -f $PIDFILE ]; then
echo -n "Shutting down $PNAME: "
killproc $PNAME
RETVAL=$?
echo
[ $RETVAL -eq 0 ] && rm -f /var/lock/subsys/$PNAME
else
echo "$PNAME does not appear to be running so will not be shut down"
fi
;;
flush)
if [ -f $PIDFILE ]; then
pid=`cat $PIDFILE`
echo Flushing buffers for $PNAME
kill -s USR1 $pid
else
echo "Can't find pid file $PIDFILE"
fi
;;
restart|reload)
$0 stop $EXT
$0 start "$2" "$3"
RETVAL=$?
;;
status)
# need to use pid file since there can be multiple instances, but only if
# this distro supports it
if [ $IFLAG -eq 1 ]; then
status -p $PIDFILE
else
status $PNAME
fi
RETVAL=$?
;;
*)
echo "Usage: $0 {start|stop|flush|restart|status}"
exit 1
esac
exit $RETVAL
collectl-3.6.9/initd/collectl-debian 0000755 0001750 0001750 00000006024 12230241727 015556 0 ustar mjs mjs #!/bin/sh
# Startup script for collectl on distros that support update-rc.d such
# as debian & ubuintu
#
### BEGIN INIT INFO
# Provides: collectl
# Required-Start: $all
# Required-Stop: $all
# Default-Start: 2 3 4 5
# Default-Stop: 0 1 6
# Short-Description: Collectl monitors system performance.
# Description: Collectl is a light-weight performance monitoring
# tool capable of reporting interactively as well
# as logging to disk. It reports statistics on
# cpu, disk, infiniband, lustre, memory, network,
# nfs, process, quadrics, slabs and more in easy
# to read format.
### END INIT INFO
# description: Run data collection for a number of subsystems
# see /etc/collectl.conf for startup options
#
# EXAMPLES:
# run at 1 second interval and only collect cpu/disk data
# /etc/init.d/collectl start "-i1 -scd"
# run a second instance with instance name of 'int5', with interval of 5 seconds
# /etc/init.d/collectl start int5 "-i5"
PERL=/usr/bin/perl
COLLECTL=/usr/bin/collectl
if [ ! -f $PERL ]; then
echo -n "Cannot find $PERL"
exit 0
fi
if [ ! -f $COLLECTL ]; then
echo -n "Cannot find $COLLECTL"
exit 0
fi
PNAME=collectl
if [ "$2" != "" ]; then
EXT=$2
if [ "$1" = "start" ] || [ "$1" = "restart" ] || [ "$1" = "force-reload" ]; then
if [ "$3" = "" ]; then
SWITCHES=$2
EXT=""
else
SWITCHES=$3
fi
fi
# Just to make sure nothing is different when running 'collectl', we
# won't use --check even though it's probably ok to use all the time.
if [ "$EXT" != "" ]; then
PNAME="collectl-$EXT"
PSWITCH="--pname $EXT"
CHECK="--check $PNAME "
fi
fi
PIDFILE="/var/run/$PNAME.pid"
case "$1" in
start)
echo -n "Starting collectl: $PNAME"
start-stop-daemon --quiet --stop --exec $PERL --pidfile $PIDFILE --test >/dev/null
if [ $? -eq 1 ]; then
start-stop-daemon --quiet --start --exec $COLLECTL -- -D $SWITCHES $PSWITCH
echo "."
else
echo " [already running]"
fi
;;
stop)
echo -n "Stopping collectl: $PNAME"
start-stop-daemon --quiet --stop --retry 2 --exec $PERL --pidfile $PIDFILE
if [ $? -eq 0 ]; then
echo "."
else
echo " [not running]"
fi
;;
flush)
start-stop-daemon --quiet --stop --exec $PERL --pidfile $PIDFILE --test >/dev/null
if [ $? -eq 0 ]; then
echo "Flushing buffers for $PNAME"
kill -s USR1 `cat $PIDFILE`
else
echo "$PNAME is not running"
fi
;;
status)
start-stop-daemon --quiet --stop --exec $PERL --pidfile $PIDFILE --test >/dev/null
if [ $? -eq 0 ]; then
echo "$PNAME is running..."
else
echo "$PNAME is not running"
exit 1
fi
;;
restart|force-reload)
$0 stop $EXT
sleep 1
$0 start "$2" "$3"
;;
*)
echo "Usage: $0 {start|stop|flush|restart|force-reload|status}"
exit 1
esac
exit 0
collectl-3.6.9/initd/collectl-generic 0000755 0001750 0001750 00000006461 12230241727 015755 0 ustar mjs mjs #!/bin/sh
# generic Startup script for collectl, in case nothing else seems to work!
#
### BEGIN INIT INFO
# Provides: collectl
# Required-Start: $all
# Required-Stop: $all
# Default-Start: 2 3 4 5
# Default-Stop: 0 1 6
# Short-Description: Collectl monitors system performance.
# Description: Collectl is a light-weight performance monitoring
# tool capable of reporting interactively as well
# as logging to disk. It reports statistics on
# cpu, disk, infiniband, lustre, memory, network,
# nfs, process, quadrics, slabs and more in easy
# to read format.
### END INIT INFO
# description: Run data collection for a number of subsystems
# see /etc/collectl.conf for startup options
# $2: process name extension, if other than collectl [optional for start/restart/reload]
# $3: parameters, in quotes if more than one, to pass to daemon command
#
# EXAMPLES:
# run at 1 second interval and only collect cpu/disk data
# /etc/init.d/collectl start "-i1 -scd"
# run a second instance with instance name of 'int5', with interval of 5 seconds
# /etc/init.d/collectl start int5 "-i5"
COLLECTL=/usr/bin/collectl
if [ ! -f $COLLECTL ]; then
echo -n "Cannot find $COLLECTL"
exit 0
fi
PNAME=collectl
if [ "$2" != "" ]; then
EXT=$2
if [ "$1" = "start" ] || [ "$1" = "restart" ] || [ "$1" = "force-reload" ]; then
if [ "$3" = "" ]; then
SWITCHES=$2
EXT=""
else
SWITCHES=$3
fi
fi
# Just to make sure nothing is different when running 'collectl', we
# won't use --check even though it's probably ok to use all the time.
if [ "$EXT" != "" ]; then
PNAME="collectl-$EXT"
PSWITCH="--pname $EXT"
CHECK="--check $PNAME "
fi
fi
PROCNAME=$PNAME
PIDFILE="/var/run/$PNAME.pid"
# If a pidfile, make sure it's not stale and if it is, collectl not running
if [ -f $PIDFILE ]; then
pid=`cat $PIDFILE`
pid=`ps ax opid,cmd | grep $PROCNAME | grep $pid | grep -v grep | awk '{ print $1 }'`
fi
case "$1" in
start)
if [ "$pid" != "" ]; then
echo "[already running]"
else
$COLLECTL -D $SWITCHES $PSWITCH
echo "."
fi
;;
stop)
if [ "$pid" = "" ]; then
echo "[not running]"
else
count=1
while [ $count -le 5 ]; do
if [ -f $PIDFILE ]; then
kill $pid
else
count=5;
break
fi
sleep 1
count=$(( $count + 1 ))
done
if [ -f $PIDFILE ]; then
echo -n " pid $pid not responding to TERM signal. sending sigkill"
kill -9 $pid
fi
echo "."
fi
;;
flush)
if [ "$pid" != "" ]; then
echo "Flushing buffers for $PNAME"
kill -s USR1 $pid
else
echo "$PNAME is not running"
fi
;;
status)
if [ "$pid" != "" ]; then
echo "$PNAME is running..."
else
echo "$PNAME is not running"
exit 1
fi
;;
restart|force-reload)
if [ "$pid" == "" ]; then
echo "$PNAME does not appear to be running so will not be shut down"
else
$0 stop $EXT
fi
$0 start "$2" "$3"
;;
*)
echo "Usage: $0 {start|stop|flush|restart|force-reload|status}"
exit 1
esac
exit 0
collectl-3.6.9/collectl.conf 0000644 0001750 0001750 00000016312 12230241726 014150 0 ustar mjs mjs # Copyright 2003-2009 Hewlett-Packard Development Company, LP
# Like most Linux configuration files, this specifies a set of user controller
# parameters. In many cases these are commented out which simply means those
# are the default values already used by collectl. To change a value
# uncomment the line and change it. To revert back to the default all you
# need do is recomment the line.
############################
# daemon/service handling
############################
# When someone specifies a daemon is to be run, typically but not limited
# to running collectl as a service, this string will cause the associated
# values to be used. They CAN also be overriden via a command line switch.
# In other words, if DaemonCommands is set to '-s cdm' and collectl is
# envoked with -D, it will process subsystems 'cdm'. However, if it is envoked
# with '-D -s mnp' it will process subsystems 'mnp', there is no combining the
# two set of values. Be sure to include any switches a daemon is required to
# have such as -f and either -r or -R.
# NOTE - if things aren't behaving as expected, you can always try running
# collectl in non-daemon mode just to see if there are any error messages. If
# you include the -m switch, you can also look in the collectl log, which is
# stored in the logging directory.
DaemonCommands = -f /var/log/collectl -r00:00,7 -m -F60 -s+YZ
# This defines the location to look for all additional required files if formatit.ph
# is not in the same directory as collectl itself.
#ReqDir = /usr/share/collectl
# E x t r a L i b r a r i e s
# So far this has only been used during development, but if there are extra
# library locations that should be 'used', put them here.
#Libraries =
# S t a n d a r d U t i l i t i e s
# Note that by default collectl will look for lspci in both /sbin and
# /usr/sbin, but if listed here will only look in that one place.
#Grep = /bin/grep
#Egrep = /bin/egrep
#Ps = /bin/ps
#Rpm = /bin/rpm
#Lspci = /sbin/lspci
#Lctl = /usr/sbin/lctl
# I n f i n i b a n d S u p p o r t
# Collectl will assume open fabric and will attempt to use the perfquery
# utility to get the counters. If not there, it assumes Voltaire and will
# first look in /proc/voltaire/adaptor-mlx/stats and failing that will use
# the get_pcounter utiliy. Since collectl resets IB counters in the
# hardware you can disable its collection by commenting out the appropriate
# variable below. PQuery for OFED, PCounter for get_pcounter calls and
# VStat for ALL non-ofed access of any kind.
# can disable either by commenting out the reference to VStat/PQuery below.
PQuery = /usr/sbin/perfquery:/usr/bin/perfquery:/usr/local/ofed/bin/perfquery
PCounter = /usr/mellanox/bin/get_pcounter
VStat = /usr/mellanox/bin/vstat:/usr/bin/vstat
OfedInfo = /usr/bin/ofed_info:/usr/local/ofed/bin/ofed_info
# Set this flag to 0 to disable checking for duplicate instances of collectl
# both trying to monitor IB stats
#IbDupCheckFlag=1
# D e f a u l t s
# This set of variables are actually all set in collectl and you need not
# change them.
# This parameter controls subsystem selection. The 'core' subsystems are
# selected when the user omits the -s switch OR uses the '+' or '-' to
# add/remove from that list. Note that changing this will also change the
# default for -s displayed in help.
#SubsysCore = bcdfijlmnstx
# although these can all be overridden by switches, they're assumed to
# always be defined so don't remove or comment any of them out! Over time
# more may be added
#Interval = 10
#Interval2 = 60
#Interval3 = 120
# These are SFS lustre specific. When using the -OD switch, any partitions
# found to be smaller than LustreSvcLunMax, which is in GB, will be ignored.
# When displaying data in verbose mode, only LustreMaxBlkSize will be
# displayed, but ALL block sizes will be read and recorded
#LustreSvcLunMax = 10
#LustreMaxBlkSize = 512
# By default, we check at these frequencies to see if lustre or interconnect
# configurations have changed. Things are efficient enough that now we can
# check for lustre changes every polling interval but I'm leaving the code
# in place rather than remove it in case needed again in the future.
#LustreConfigInt = 1
#InterConnectInt = 900
# These apply to disk/partition limits for exception (-o x/X) processing
#LimSVC = 30 # Minimum partition Avg Service time
#LimIOS = 10 # Minumum number of Disk OR Partion I/Os
#LimBool = 0 # generate exception record if EITHER limit exceeded
#LimLusKBS = 100 # Minimum number of Lustre OSS KB/sec
#LimLusReints = 1000 # Minimum number o Lustre MDS Reint operations
# Socket I/O Defaults
#Port = 2655
#Timeout = 10
# Maximum allowable zlib errors in a single day or run.
#MaxZlibErrors = 20
# In order to detect bogus network data we need to know what is normal for that
# interface. Collectl uses ethtool for 'eth' interfaces to get the default
# speed but that requires elevated privileges. It the ethtool fails, this
# value (which is in Mb) will be used instead.
# To disable bogus network data checking, set this to any negative value
#DefNetSpeed=10000
# Collectl will automatically size the frequency of headers in 'brief format'
# to the height of your display window which it determines using the resize
# utility. If that utility can't be found, it will use the height speficied
# in 'TermSize'. If 'resize' is in your path but you want a fixed/different
# size, comment out the Resize line and uncomment TermHeight, setting it to
# what you want.
#TermHeight = 24
Resize=/usr/bin/resize:/usr/X11R6/bin/resize
# To turn off Time:HiRes/glibc incompatibility checking, the following
# should be enabled and set to 0
#TimeHiResCheck = 1
# These control environmental monitoring and to use it you MUST have ipmitool
# installed (see http://ipmitool.sourceforge.net/). If not in the path shown
# below, you must change it.
Ipmitool = /usr/bin/ipmitool:/usr/local/bin/ipmitool:/opt/hptc/sbin/ipmitool
IpmiCache = /var/run/collectl-ipmicache
IpmiTypes = fan,temp,current
# passwd file for UID to usernames mapping during process monitoring
#Passwd = /etc/passwd
# If a cciss device is reset (such as when during a lun scan) while collectl running,
# disk rates will be excessive. If one seen above the following, reset ALL stats for
# that disk to 0. To disable set this to -1
#DiskMaxValue=5000000
# When collectl reads disk data, it filters out any that don't match the DiskFilter,
# which by default looks for cciss, hd, sd, xvd, dm, emcpower and psv. All others are
# ignored. To change the filter, set the string below to those you want to keep BUT
# you need to know what a perl regular expression looks like or you may not get the
# desired results. CAUTION - white space is CRITICAL for this to work.
#DiskFilter = /cciss/c\d+d\d+ |hd[ab] | sd[a-z]+ |dm-\d+ |xvd[a-z] |fio[a-z]+ | vd[a-z]+ |emcpower[a-z]+ |psv\d+ /
# Kernel Efficiency Test
# On kernels 2.6.32 forward (and you can't tell how distros patched) there is a read inefficiency
# in the /proc filesystem for 4 and more sockets and the only way to tell is to test it. If slow
# generate a warning that patching the kernel may be recommmended. To bypass the test/message, set
# the following to 'no'
#ProcReadTest = yes
collectl-3.6.9/INSTALL 0000755 0001750 0001750 00000006137 12230241726 012540 0 ustar mjs mjs #!/bin/sh
DESTDIR=${DESTDIR:="/"}
BINDIR=$DESTDIR/usr/bin
DOCDIR=$DESTDIR/usr/share/doc/collectl
SHRDIR=$DESTDIR/usr/share/collectl
MANDIR=$DESTDIR/usr/share/man/man1
ETCDIR=$DESTDIR/etc
INITDIR=$ETCDIR/init.d
mkdir -p $BINDIR
mkdir -p $DOCDIR
mkdir -p $SHRDIR
mkdir -p $ETCDIR
mkdir -p $MANDIR
mkdir -p $INITDIR
mkdir -p $SHRDIR/util
mkdir -p $DESTDIR/var/log/collectl
cp collectl.pl $BINDIR/collectl
cp collectl.conf $ETCDIR
cp man1/* $MANDIR
cp initd/* $INITDIR
cp docs/* $DOCDIR
cp GPL ARTISTIC COPYING $DOCDIR
cp RELEASE-collectl $DOCDIR
cp UNINSTALL $SHRDIR
cp formatit.ph $SHRDIR
cp lexpr.ph sexpr.ph $SHRDIR
cp gexpr.ph misc.ph $SHRDIR
cp hello.ph graphite.ph $SHRDIR
cp envrules.std $SHRDIR
cp vmstat.ph $SHRDIR
cp client.pl $SHRDIR/util
cp col2tlviz.pl $SHRDIR/util
# Force in case redoing the install and files already zipped
gzip -f $MANDIR/collectl*
chmod 755 $INITDIR/collectl*
chmod 444 $ETCDIR/collectl.conf
chmod 755 $BINDIR/collectl
chmod 444 $DOCDIR/ARTISTIC $DOCDIR/COPYING $DOCDIR/GPL
chmod 755 $SHRDIR/util/*
chmod 444 $SHRDIR/formatit.ph $SHRDIR/*expr.ph $SHRDIR/vmstat.ph
# remove any stale versions in case the names/numbers used have changed.
# on new ROCKS installion 'rm' isn't there yet! [thanks roy]
if [ -x /bin/rm ] ; then
/bin/rm -f $INITDIR/rc*.d/*collectl
/bin/rm -f $ETCDIR/rc.d/rc*.d/*collectl
fi
# Try and decide which distro this is based on distro specific files.
distro=1
if [ -f /sbin/yast ]; then
distro=2
mv -f $INITDIR/collectl-suse $INITDIR/collectl
rm -f $INITDIR/collectl-debian
rm -f $INITDIR/collectl-generic
fi
# debian
if [ -f /usr/sbin/update-rc.d ]; then
distro=3
mv -f $INITDIR/collectl-debian $INITDIR/collectl
rm -f $INITDIR/collectl-suse
rm -f $INITDIR/collectl-generic
# only if we're installing under /
[ "$DESTDIR" = "/" ] && update-rc.d collectl defaults
fi
# redhat
if [ -f /etc/redhat-release ]; then
distro=4
rm -f $INITDIR/collectl-suse
rm -f $INITDIR/collectl-debian
rm -f $INITDIR/collectl-generic
[ "$DESTDIR" = "/" ] && chkconfig --add collectl
fi
# gentoo
if [ -f $ETCDIR/gentoo-release ]; then
distro=5
mv -f $INITDIR/collectl-generic $INITDIR/collectl
rm -f $INITDIR/collectl-suse
rm -f $INITDIR/collectl-debian
[ "$DESTDIR" = "/" ] && rc-update -a collectl default
fi
# Generic Distros
# If /etc/init.d doesn't exist and/or there's no way to use chkconfig or
# rc-update you're going to have to add some custom code below...
if [ ${distro} = 1 ]; then
mv -f $INITDIR/collectl-generic $INITDIR/collectl
rm -f $INITDIR/collectl-suse
rm -f $INITDIR/collectl-debian
# If in not installing under / there's nothing extra do
[ $DESTDIR != "/" ] && exit 0
# figure out how to handle reboots
if [ -f /sbin/chkconfig ]; then
chkconfig --add collectl
elif [ -f /sbin/rc-update ]; then
rc-update -a collectl default
else
echo "could not figure out how to enable restarting across reboots"
fi
fi
collectl-3.6.9/graphite.ph 0000644 0001750 0001750 00000057076 12230241726 013650 0 ustar mjs mjs # copyright, 2003-20012 Hewlett-Packard Development Company, LP
# debug
# 1 - print Var, Units and Values
# 2 - only print sent 'changed' Var/Units/Vales
# 4 - not used
# 8 - do not open/use socket (typically used with other flags)
# 16 - print socket open/close info
our $graphiteInterval;
my $graphiteTimeout=5;
my $graphiteCounter=0;
my $graphiteSocketFailMax=5; # report socket open fails every 100 intervals
my $graphiteIntTimeLast=0; # tracks start of new interval
my $graphiteOneTB=1024*1024*1024*1024;
my $graphitePost=''; # insert AFTER hostname in message to carbon (don't forget '.' if you want one)
my ($graphiteSubsys, $graphiteBefore, $graphiteEscape, $graphiteRandomize);
my ($graphiteDebug, $graphiteColInt, $graphiteCOFlag, $graphiteSendCount);
my ($graphiteTTL, %graphiteTTL, %graphiteDataMin, %graphiteDataMax, %graphiteDataTot, %graphiteDataLast);
my ($graphiteMyHost, $graphiteSocket, $graphiteSockHost, $graphiteSockPort, $graphiteSocketFailCount);
my ($graphiteAlign, $graphiteFqdnFlag, $graphiteMinFlag, $graphiteMaxFlag, $graphiteAvgFlag, $graphiteTotFlag, $graphiteFlags)=(0,0,0,0,0,0,0);
my $graphiteOutputFlag=1;
# This sets a flag as soon as we 'require' the module and tells collectl this
# module does socket communications w/o -A and so is ok to run as daemon
# without requiring -f or -A.
$exportComm |= 1;
sub graphiteInit
{
my $hostport=shift;
help() if $hostport eq 'h';
error("host[:port] must be specified as first parameter") if !defined($hostport);
error('--showcolheader not supported by graphite') if $showColFlag;
# Just like vmstat
error("-f requires either --rawtoo or -P") if $filename ne '' && !$rawtooFlag && !$plotFlag;
error("-P or --rawtoo require -f") if $filename eq '' && ($rawtooFlag || $plotFlag);
# parameter defaults
$hostport.=":2003" if $hostport!~/:/;
$graphiteDebug=$graphiteCOFlag=0;
$graphiteInterval='';
$graphiteSubsys=$subsys;
$graphiteTTL=5;
$graphiteBefore='';
$graphiteEscape='';
$graphiteRandomize='';
foreach my $option (@_)
{
my ($name, $value)=split(/=/, $option);
error("invalid graphite option '$name'") if $name!~/^[bdefhiprs]?$|^align|^co$|^ttl$|^min$|^max$|^avg$|^tot$/;
$graphiteAlignFlag=1 if $name eq 'align';
$graphiteBefore=$value if $name eq 'b';
$graphiteCOFlag=1 if $name eq 'co';
$graphiteDebug=$value if $name eq 'd';
$graphiteEscape=$value if $name eq 'e';
$graphiteInterval=$value if $name eq 'i';
$graphitePost=$value if $name eq 'p';
$graphiteRandomize=$value if $name eq 'r';
$graphiteSubsys=$value if $name eq 's';
$graphiteTTL=$value if $name eq 'ttl';
$graphiteFqdnFlag=1 if $name eq 'f';
$graphiteMinFlag=1 if $name eq 'min';
$graphiteMaxFlag=1 if $name eq 'max';
$graphiteAvgFlag=1 if $name eq 'avg';
$graphiteTotFlag=1 if $name eq 'tot';
help() if $name eq 'h';
}
error("graphite does not support standard collectl socket I/O via -A") if $graphiteSockFlag;
($graphiteSockHost, $graphiteSockPort)=split(/:/, $hostport);
error("the port number must be specified") if !defined($graphiteSockPort) || $graphiteSockPort eq '';
# s= disables ALL subsys, only makes sense with imports
error("graphite subsys options '$graphiteSubsys' not a proper subset of '$subsys'")
if $subsys ne '' && $graphiteSubsys ne '' && $graphiteSubsys!~/^[$subsys]+$/;
$graphiteFlags=$graphiteMinFlag+$graphiteMaxFlag+$graphiteAvgFlag+$graphiteTotFlag;
error("only 1 of 'min', 'max', 'avg' or 'tot' with 'graphite'") if $graphiteFlags>1;
# If we ever run with a ':' in the interval, we need to be sure we're only looking at the main one.
$graphiteColInt=(split(/:/, $interval))[0];
$graphiteInterval=$graphiteColInt if $graphiteInterval eq '';
# convert to the number of samples we want to send
$graphiteSendCount=int($graphiteInterval/$graphiteColInt);
error("graphite interval '$graphiteInterval' is not a multiple of '$graphiteColInt' seconds")
if $graphiteColInt*$graphiteSendCount != $graphiteInterval;
error("'min', 'max', 'avg' & 'tot' require graphite 'i' that is > collectl's -i")
if $graphiteFlags && $graphiteSendCount==1;
if ($graphiteAlignFlag)
{
my $div1=int(60/$graphiteColInt);
my $div2=int($graphiteColInt/60);
error("'align' requires collectl interval be a factor or multiple of 60 seconds")
if ($graphiteColInt<=60 && $div1*$graphiteColInt!=60) || ($graphiteColInt>60 && $div2*60!=$graphiteColInt);
error("'align' only makes sense when multiple samples/interval") if $graphiteInterval<=$graphiteColInt;
error("'lexpr,align' requires -D or --align") if !$graphiteAlignFlag && !$daemonFlag;
}
error('randomize options requires a value') if !defined($graphiteRandomize);
if ($graphiteRandomize ne '')
{
error("randomization require hires time module") if !$hiResFlag;
error("randomization requires interval of at least 2 seconds") if $graphiteInterval<2;
error("randomization value must be less than or equal to '" . ($graphiteInterval-1) . "' seconds")
if $graphiteRandomize > $graphiteInterval-1;
}
# Since graphite DOES write over a socket but does not use -A, make sure the default
# behavior for -f logs matches that of -A
$rawtooFlag=1 if $filename ne '' && !$plotFlag;
$graphiteMyHost=(!$graphiteFqdnFlag) ? `hostname` : `hostname -f`;
chomp $graphiteMyHost;
$graphiteMyHost =~ s/\./$graphiteEscape/g if $graphiteEscape ne '';
# O p e n S o c k e t
$SIG{"PIPE"}=\&graphiteSigPipe; # socket comm errors
# set fail count such that if first open fails, we'll report an error
$graphiteSocketFailCount=$graphiteSocketFailMax-1;
openTcpSocket(1);
}
# NOTE - this routine is almost an identical copy from gexpr.
# Being lazy while making it easier to keep the 2 in sync, I left in the
# second parameter in the sendData() calls which are ignored in the
# modified version of sendData() itself, which prepends a hostname to the
# variable name and add a timestamp to the socket call. In fact, I almost
# just hacked up gexpr to make it deal with both ganglia and graphite.
sub graphite
{
# if socket not even open and the first try of this interval, try again
# NOTE - we're making sure socket is open every interval whether we're
# reporting data or not...
openTcpSocket() if !defined($graphiteSocket) && $graphiteIntTimeLast!=time;
$graphiteIntTimeLast=time;
return if !defined($graphiteSocket) && !($graphiteDebug & 8); # still not open? get out!
# if not time to print and we're not doing min/max/avg/tot, there's nothing to do.
# BUT always make sure time aligns to top of minute based on i=
$graphiteCounter++;
$graphiteOutputFlag=(($graphiteCounter % $graphiteSendCount) == 0) ? 1 : 0 if !$graphiteAlignFlag;
$graphiteOutputFlag=(!(int($lastSecs[$rawPFlag]) % $graphiteInterval)) ? 1 : 0 if $graphiteAlignFlag;
return if (!$graphiteOutputFlag && $graphiteFlags==0);
# random sleep when r= option
Time::HiRes::usleep(rand($graphiteRandomize)*1000000) if $graphiteRandomize ne '';
if ($graphiteSubsys=~/c/)
{
# CPU utilization is a % and we don't want to report fractions
my $i=$NumCpus;
sendData('cputotals.user', 'percent', $userP[$i]);
sendData('cputotals.nice', 'percent', $niceP[$i]);
sendData('cputotals.sys', 'percent', $sysP[$i]);
sendData('cputotals.wait', 'percent', $waitP[$i]);
sendData('cputotals.idle', 'percent', $idleP[$i]);
sendData('cputotals.irq', 'percent', $irqP[$i]);
sendData('cputotals.soft', 'percent', $softP[$i]);
sendData('cputotals.steal','percent', $stealP[$i]);
sendData('ctxint.ctx', 'switches/sec', $ctxt/$intSecs);
sendData('ctxint.int', 'intrpts/sec', $intrpt/$intSecs);
sendData('ctxint.proc', 'pcreates/sec', $proc/$intSecs);
sendData('ctxint.runq', 'runqSize', $loadQue);
# these are the ONLY fraction, noting they will print to 2 decimal places
sendData('cpuload.avg1', 'loadAvg1', $loadAvg1, 2);
sendData('cpuload.avg5', 'loadAvg5', $loadAvg5, 2);
sendData('cpuload.avg15', 'loadAvg15', $loadAvg15, 2);
}
if ($graphiteSubsys=~/C/)
{
for (my $i=0; $i<$NumCpus; $i++)
{
sendData("cpuinfo.user.cpu$i", 'percent', $userP[$i]);
sendData("cpuinfo.nice.cpu$i", 'percent', $niceP[$i]);
sendData("cpuinfo.sys.cpu$i", 'percent', $sysP[$i]);
sendData("cpuinfo.wait.cpu$i", 'percent', $waitP[$i]);
sendData("cpuinfo.irq.cpu$i", 'percent', $irqP[$i]);
sendData("cpuinfo.soft.cpu$i", 'percent', $softP[$i]);
sendData("cpuinfo.steal.cpu$i", 'percent', $stealP[$i]);
sendData("cpuinfo.idle.cpu$i", 'percent', $idleP[$i]);
sendData("cpuinfo.intrpt.cpu$i",'percent', $intrptTot[$i]);
}
}
if ($graphiteSubsys=~/d/)
{
sendData('disktotals.reads', 'reads/sec', $dskReadTot/$intSecs);
sendData('disktotals.readkbs', 'readkbs/sec', $dskReadKBTot/$intSecs);
sendData('disktotals.writes', 'writes/sec', $dskWriteTot/$intSecs);
sendData('disktotals.writekbs', 'writekbs/sec', $dskWriteKBTot/$intSecs);
}
if ($graphiteSubsys=~/D/)
{
for (my $i=0; $i<@dskOrder; $i++)
{
# preserve display order but skip any disks not seen this interval
$dskName=$dskOrder[$i];
next if !defined($dskSeen[$i]);
next if ($dskFiltKeep eq '' && $dskName=~/$dskFiltIgnore/) || ($dskFiltKeep ne '' && $dskName!~/$dskFiltKeep/);
sendData("diskinfo.reads.$dskName", 'reads/sec', $dskRead[$i]/$intSecs);
sendData("diskinfo.readkbs.$dskName", 'readkbs/sec', $dskReadKB[$i]/$intSecs);
sendData("diskinfo.writes.$dskName", 'writes/sec', $dskWrite[$i]/$intSecs);
sendData("diskinfo.writekbs.$dskName", 'writekbs/sec', $dskWriteKB[$i]/$intSecs);
}
}
if ($graphiteSubsys=~/f/)
{
if ($nfsSFlag)
{
sendData('nfsinfo.SRead', 'SvrReads/sec', $nfsSReadsTot/$intSecs);
sendData('nfsinfo.SWrite', 'SvrWrites/sec', $nfsSWritesTot/$intSecs);
sendData('nfsinfo.Smeta', 'SvrMeta/sec', $nfsSMetaTot/$intSecs);
sendData('nfsinfo.Scommit', 'SvrCommt/sec' , $nfsSCommitTot/$intSecs);
}
if ($nfsCFlag)
{
sendData('nfsinfo.CRead', 'CltReads/sec', $nfsCReadsTot/$intSecs);
sendData('nfsinfo.CWrite', 'CltWrites/sec', $nfsCWritesTot/$intSecs);
sendData('nfsinfo.Cmeta', 'CltMeta/sec', $nfsCMetaTot/$intSecs);
sendData('nfsinfo.Ccommit', 'CltCommt/sec' , $nfsCCommitTot/$intSecs);
}
}
if ($graphiteSubsys=~/i/)
{
sendData('inodeinfo.dentnum', 'dentrynum', $dentryNum);
sendData('inodeinfo.dentunused', 'dentryunused', $dentryUnused);
sendData('inodeinfo.fhandalloc', 'filesalloc', $filesAlloc);
sendData('inodeinfo.fhandmpct', 'filesmax', $filesMax);
sendData('inodeinfo.inodenum', 'inodeused', $inodeUsed);
}
if ($graphiteSubsys=~/l/)
{
if ($CltFlag)
{
sendData('lusclt.reads', 'reads/sec', $lustreCltReadTot/$intSecs);
sendData('lusclt.readkbs', 'readkbs/sec', $lustreCltReadKBTot/$intSecs);
sendData('lusclt.writes', 'writes/sec', $lustreCltWriteTot/$intSecs);
sendData('lusclt.writekbs', 'writekbs/sec', $lustreCltWriteKBTot/$intSecs);
sendData('lusclt.numfs', 'filesystems', $NumLustreFS);
}
if ($MdsFlag)
{
my $getattrPlus=$lustreMdsGetattr+$lustreMdsGetattrLock+$lustreMdsGetxattr;
my $setattrPlus=$lustreMdsReintSetattr+$lustreMdsSetxattr;
my $varName=($cfsVersion lt '1.6.5') ? 'reint' : 'unlink';
my $varVal= ($cfsVersion lt '1.6.5') ? $lustreMdsReint : $lustreMdsReintUnlink;
sendData('lusmds.gattrP', 'gattrP/sec', $getattrPlus/$intSecs);
sendData('lusmds.sattrP', 'sattrP/sec', $setattrPlus/$intSecs);
sendData('lusmds.sync', 'sync/sec', $lustreMdsSync/$intSecs);
sendData("lusmds.$varName", "$varName/sec", $varVal/$intSecs);
}
if ($OstFlag)
{
sendData('lusost.reads', 'reads/sec', $lustreReadOpsTot/$intSecs);
sendData('lusost.readkbs', 'readkbs/sec', $lustreReadKBytesTot/$intSecs);
sendData('lusost.writes', 'writes/sec', $lustreWriteOpsTot/$intSecs);
sendData('lusost.writekbs', 'writekbs/sec', $lustreWriteKBytesTot/$intSecs);
}
}
if ($graphiteSubsys=~/L/)
{
if ($CltFlag)
{
# Either report details by filesystem OR OST
if ($lustOpts!~/O/)
{
for (my $i=0; $i<$NumLustreFS; $i++)
{
sendData("lusost.reads.$lustreCltFS[$i]", 'reads/sec', $lustreCltRead[$i]/$intSecs);
sendData("lusost.readkbs.$lustreCltFS[$i]", 'readkbs/sec', $lustreCltReadKB[$i]/$intSecs);
sendData("lusost.writes.$lustreCltFS[$i]", 'writes/sec', $lustreCltWrite[$i]/$intSecs);
sendData("lusost.writekbs.$lustreCltFS[$i]", 'writekbs/sec', $lustreCltWriteKB[$i]/$intSecs);
}
}
else
{
for (my $i=0; $i<$NumLustreCltOsts; $i++)
{
sendData("lusost.reads.$lustreCltOsts[$i]", 'reads/sec', $lustreCltLunRead[$i]/$intSecs);
sendData("lusost.readkbs.$lustreCltOsts[$i]", 'readkbs/sec', $lustreCltLunReadKB[$i]/$intSecs);
sendData("lusost.writes.$lustreCltOsts[$i]", 'writes/sec', $lustreCltLunWrite[$i]/$intSecs);
sendData("lusost.writekbs.$lustreCltOsts[$i]", 'writekbs/sec', $lustreCltLunWriteKB[$i]/$intSecs);
}
}
}
if ($OstFlag)
{
for ($i=0; $i<$NumOst; $i++)
{
sendData("lusost.reads.$lustreOsts[$i]", 'reads/sec', $lustreReadOps[$i]/$intSecs);
sendData("lusost.readkbs.$lustreOsts[$i]", 'readkbs/sec', $lustreReadKBytes[$i]/$intSecs);
sendData("lusost.writes.$lustreOsts[$i]", 'writes/sec', $lustreWriteOps[$i]/$intSecs);
sendData("lusost.writekbs.$lustreOsts[$i]", 'writekbs/sec', $lustreWriteKBytes[$i]/$intSecs);
}
}
}
if ($graphiteSubsys=~/m/)
{
sendData('meminfo.tot', 'kb', $memTot);
sendData('meminfo.free', 'kb', $memFree);
sendData('meminfo.shared', 'kb', $memShared);
sendData('meminfo.buf', 'kb', $memBuf);
sendData('meminfo.cached', 'kb', $memCached);
sendData('meminfo.used', 'kb', $memUsed);
sendData('meminfo.slab', 'kb', $memSlab);
sendData('meminfo.map', 'kb', $memMap);
sendData('meminfo.hugetot', 'kb', $memHugeTot);
sendData('meminfo.hugefree', 'kb', $memHugeFree);
sendData('meminfo.hugersvd', 'kb', $memHugeRsvd);
sendData('swapinfo.total', 'kb', $swapTotal);
sendData('swapinfo.free', 'kb', $swapFree);
sendData('swapinfo.used', 'kb', $swapUsed);
sendData('swapinfo.in', 'swaps/sec', $swapin/$intSecs);
sendData('swapinfo.out', 'swaps/sec', $swapout/$intSecs);
sendData('pageinfo.fault', 'faults/sec', $pagefault/$intSecs);
sendData('pageinfo.majfault', 'majflt/sec', $pagemajfault/$intSecs);
sendData('pageinfo.in', 'pages/sec', $pagein/$intSecs);
sendData('pageinfo.out', 'pages/sec', $pageout/$intSecs);
}
if ($graphiteSubsys=~/M/)
{
for (my $i=0; $i<$CpuNodes; $i++)
{
foreach my $field ('used', 'free', 'slab', 'map', 'anon', 'lock', 'act', 'inact')
{
sendData("numainfo.$field.$i", 'kb', $numaMem[$i]->{$field});
}
}
}
if ($graphiteSubsys=~/n/)
{
sendData('nettotals.kbin', 'kb/sec', $netRxKBTot/$intSecs);
sendData('nettotals.pktin', 'kb/sec', $netRxPktTot/$intSecs);
sendData('nettotals.kbout', 'kb/sec', $netTxKBTot/$intSecs);
sendData('nettotals.pktout', 'kb/sec', $netTxPktTot/$intSecs);
}
if ($graphiteSubsys=~/N/)
{
for ($i=0; $i<@netOrder; $i++)
{
$netName=$netOrder[$i];
next if !defined($netSeen[$i]);
next if ($netFiltKeep eq '' && $netName=~/$netFiltIgnore/) || ($netFiltKeep ne '' && $netName!~/$netFiltKeep/);
next if $netName=~/lo|sit/;
sendData("nettotals.kbin.$netName", 'kb/sec', $netRxKB[$i]/$intSecs);
sendData("nettotals.pktin.$netName", 'kb/sec', $netRxPkt[$i]/$intSecs);
sendData("nettotals.kbout.$netName", 'kb/sec', $netTxKB[$i]/$intSecs);
sendData("nettotals.pktout.$netName", 'kb/sec', $netTxPkt[$i]/$intSecs);
}
}
if ($graphiteSubsys=~/s/)
{
sendData("sockinfo.used", 'sockets', $sockUsed);
sendData("sockinfo.tcp", 'sockets', $sockTcp);
sendData("sockinfo.orphan",'sockets', $sockOrphan);
sendData("sockinfo.tw", 'sockets', $sockTw);
sendData("sockinfo.alloc", 'sockets', $sockAlloc);
sendData("sockinfo.mem", 'sockets', $sockMem);
sendData("sockinfo.udp", 'sockets', $sockUdp);
sendData("sockinfo.raw", 'sockets', $sockRaw);
sendData("sockinfo.frag", 'sockets', $sockFrag);
sendData("sockinfo.fragm", 'sockets', $sockFragM);
}
if ($graphiteSubsys=~/t/)
{
sendData("tcpinfo.iperrs", 'num/sec', $ipErrors/$intSecs) if $tcpFilt=~/i/;
sendData("tcpinfo.tcperrs", 'num/sec', $tcpErrors/$intSecs) if $tcpFilt=~/t/;
sendData("tcpinfo.udperrs", 'num/sec', $udpErrors/$intSecs) if $tcpFilt=~/u/;
sendData("tcpinfo.icmperrs", 'num/sec', $icmpErrors/$intSecs) if $tcpFilt=~/c/;
sendData("tcpinfo.tcpxerrs", 'num/sec', $tcpExErrors/$intSecs) if $tcpFilt=~/T/;
}
if ($graphiteSubsys=~/x/i)
{
if ($NumXRails)
{
$kbInT= $elanRxKBTot;
$pktInT= $elanRxTot;
$kbOutT= $elanTxKBTot;
$pktOutT=$elanTxTot;
}
if ($NumHCAs)
{
$kbInT= $ibRxKBTot;
$pktInT= $ibRxTot;
$kbOutT= $ibTxKBTot;
$pktOutT=$ibTxTot;
}
sendData("iconnect.kbin", 'kb/sec', $kbInT/$intSecs);
sendData("iconnect.pktin", 'pkt/sec', $pktInT/$intSecs);
sendData("iconnect.kbout", 'kb/sec', $kbOutT/$intSecs);
sendData("iconnect.pktout", 'pkt/sec', $pktOutT/$intSecs);
}
if ($graphiteSubsys=~/E/i)
{
foreach $key (sort keys %$ipmiData)
{
for (my $i=0; $i{$key}}); $i++)
{
my $name=$ipmiData->{$key}->[$i]->{name};
my $inst=($key!~/power/ && $ipmiData->{$key}->[$i]->{inst} ne '-1') ? $ipmiData->{$key}->[$i]->{inst} : '';
sendData("env.$name$inst", $name, $ipmiData->{$key}->[$i]->{value}, '%s');
}
}
}
my (@names, @units, @vals);
for (my $i=0; $i<$impNumMods; $i++) { &{$impPrintExport[$i]}('g', \@names, \@units, \@vals); }
foreach (my $i=0; $i$graphiteDataMax{$name};
$graphiteDataTot{$name}+=$value if $graphiteAvgFlag || $graphiteTotFlag;
}
return('') if !$graphiteOutputFlag;
# A c t u a l S e n d H a p p e n s H e r e
# If doing min/max/avg, reset $value
if ($graphiteFlags)
{
$value=$graphiteDataMin{$name} if $graphiteMinFlag;
$value=$graphiteDataMax{$name} if $graphiteMaxFlag;
$value=$graphiteDataTot{$name} if $graphiteTotFlag;
$value=($graphiteDataTot{$name}/$graphiteCounter) if $graphiteAvgFlag;
}
# Always send send data if not CO mode, but if so only send when it has
# indeed changed OR TTL about to expire
my $valSentFlag=0;
if (!$graphiteCOFlag || $value!=$graphiteDataLast{$name} || $graphiteTTL{$name}==1)
{
$valSentFlag=1;
my $valString=(!defined($numpl)) ? sprintf('%d', $value) : sprintf("%.${numpl}f", $value);
my $message=sprintf("$graphiteBefore$graphiteMyHost$graphitePost.$name $valString %d\n", $graphiteIntTimeLast);
print $message if $graphiteDebug & 1;
if (!($graphiteDebug & 8))
{
my $bytes=syswrite($graphiteSocket, $message, length($message), 0);
}
$graphiteDataLast{$name}=$value;
}
# TTL only applies when in 'CO' mode
if ($graphiteCOFlag)
{
$graphiteTTL{$name}-- if !$valSentFlag;
$graphiteTTL{$name}=$graphiteTTL if $valSentFlag || $graphiteTTL{$name}==0;
}
}
sub openTcpSocket
{
return if $graphiteDebug & 8; # don't open socket
print "Opening Socket on $graphiteSockHost:$graphiteSockPort\n" if $graphiteDebug & 16;
$graphiteSocket=new IO::Socket::INET(
PeerAddr => $graphiteSockHost,
PeerPort => $graphiteSockPort,
Proto => 'tcp',
Timeout => $graphiteTimeout);
if (!defined($graphiteSocket))
{
if (++$graphiteSocketFailCount==$graphiteSocketFailMax)
{
logmsg('E', "Could not create socket to $graphiteSockHost:$graphiteSockPort. Reason: $!");
$graphiteSocketFailCount=0;
}
}
else
{
# we're printing to the term with d=16 because 'I' messages don't go there.
my $message="Socket opened to graphite/carbon on $graphiteSockHost:$graphiteSockPort";
print "$message\n" if $graphiteDebug & 16;
logmsg('I', $message);
$graphiteSocketFailCount=0;
}
}
# This catches the socket failure. Only problem is it doesn't happen until we try write
# and as a result when we return the write fails with an undef on the socket variable.
# Not really a big deal...
sub graphiteSigPipe
{
undef $graphiteSocket;
}
sub help
{
my $text=<[$kids]=$pidZ;
#printf "PID: $pidZ I: $i PPID: $ppidZ KIDS: $kids\n";
}
if ($debug & 1)
{
# Look for orphans. Probably only useful for debugging and then not even sure...
foreach my $pid (keys %procIndexes)
{
my $i=$procIndexes{$pid};
my $ppid=(defined($procTgid[$i]) && $procTgid[$i]!=$procPid[$i]) ? $procTgid[$i] : $procPpid[$i];
my $ppidZ=sprintf("%05d", $ppid);
print "*** Pid $pid is an orphan ***\n" if !defined($procChild{$ppidZ});
}
foreach my $ppid (sort keys %procChild)
{
printf "Parent: %5d Kids: %5d PIDS:", $ppid, scalar(@{$procChild{$ppid}});
foreach my $pid (@{$procChild{$ppid}})
{
printf " %d", $pid;
}
print "\n";
}
}
printTreeHeader();
aggregate('00000') if $aggregateFlag;
$procCount=0;
$level{'00000'}=-1;
push @stack, '00000';
my $i2=$interval2Secs;
while (scalar(@stack))
{
my $ppidZ=pop(@stack);
my $i=$procIndexes{$ppidZ*1};
#print "POPPED: $ppidZ I: $i LEVEL: $level{$ppidZ}\n";
my $level=$level{$ppidZ}+1;
foreach my $pidZ (@{$procChild{$ppidZ}})
{
$level{$pidZ}=$level;
push @stack, $pidZ;
}
next if $level==0 || $level>$depth;
next if !$threadsFlag && $procThread[$i];
if ($skipFlag)
{
next if $topType eq 'syst' && $procSTime[$i]==0;
next if $topType eq 'usrt' && $procUTime[$i]==0;
next if $topType eq 'time' && $procSTime[$i]+$procUTime[$i]==0;
if ($procOpts!~/f/)
{
next if $topType eq 'majf' && $procMajFlt[$i]==0;
next if $topType eq 'minf' && $procMinFlt[$i]==0;
next if $topType eq 'flt' && $procMajFlt[$i]+$procMinFlt[$i]==0;
}
else
{
next if $topType eq 'majf' && $procMajFltTot[$i]==0;
next if $topType eq 'minf' && $procMinFltTot[$i]==0;
next if $topType eq 'flt' && $procMajFltTot[$i]+$procMinFltTot[$i]==0;
}
# I/O KBs are special...
next if $topType eq 'rkb' && $procRKB[$i]/$i2<=$ioSkip;
next if $topType eq 'wkb' && $procWKB[$i]/$i2<=$ioSkip;
next if $topType eq 'iokb' && ($procRKB[$i]+$procWKB[$i])/$i2<=$ioSkip;
next if $topType eq 'rkbc' && $procRKBC[$i]/$i2<=$ioSkip;
next if $topType eq 'wkbc' && $procWKBC[$i]/$i2<=$ioSkip;
next if $topType eq 'iokbc' && ($procRKBC[$i]+$procWKBC[$i])/$i2<=$ioSkip;
next if $topType eq 'ioall' && ($procRKB[$i]+$procWKB[$i]+$procRKBC[$i]+$procWKBC[$i])/$i2<=$ioSkip;
next if $topType eq 'rsys' && $procRSys[$i]==0;
next if $topType eq 'wsys' && $procWSys[$i]==0;
next if $topType eq 'iosys' && $procRSys[$i]+$procWSys[$i]==0;
}
next if defined($pidSelect) && ($ppidZ ne $pidSelect) && !$pidPrinted;
$pidPrinted=1;
my $parent=defined($procTgid[$i]) && $procTgid[$i]!=$procPid[$i] ? $procTgid[$i] : $procPpid[$i];
my $pIndex=$procIndexes{$parent};
printPid($pIndex, $parent) if $procThread[$i] && !defined($procPrinted{$parent});
printPid($i, $ppidZ*1);
last if $numTop && $procCount>=$numTop;
}
$clscr=$home;
print $clr;
}
sub printTreeHeader
{
my $tempTime='';
$tempTime= " ".(split(/\s+/,localtime($lastInt2Secs)))[3];
$tempTime.=sprintf(".%03d", $usecs) if $options=~/m/;
my $line="Process Tree$tempTime ";
$line.=sprintf("[skip when '$expSkip'<=%d%s is '%s' ", $ioSkip, $ioSkip ? 'KB' : '', $skipFlag ? 'on' : 'off');
$line.=sprintf("aggr: '%s' x1024: '%s' depth $depth",
$aggregateFlag ? 'on' : 'off', $kFlag ? 'on' : 'off');
$line.=sprintf(" threads: %s", $threadsFlag ? 'on' : 'off') if $procOpts=~/t/;
$line.="]$eol\n";
$line.=$eol if $playback eq '' && $numTop;
printText("\n") if !$homeFlag;
printText($line);
my $filler=' 'x$depth;
if ($procOpts!~/[im]/)
{
$tempHdr= "# PID$filler PPID User PR S VSZ RSS CP SysT UsrT Pct AccuTime ";
$tempHdr.=" RKB WKB " if $processIOFlag;
$tempHdr.="MajF MinF Command\n";
}
elsif ($procOpts=~/i/)
{
$tempHdr= "# PID$filler PPID User S SysT UsrT AccuTime RKB WKB RKBC WKBC RSys WSys Cncl Command\n";
}
elsif ($procOpts=~/m/)
{
$tempHdr= "# PID$filler PPID User S VmSize VmLck VmRSS VmData VmStk VmExe VmLib MajF MinF Command\n";
}
printText($tempHdr);
}
sub printPid
{
my $i= shift;
my $ppid=shift;
my $ppidZ=sprintf("%05d", $ppid);
my $pad=($level{$ppidZ});
my $padL=' 'x$pad;
my $padR=' 'x($depth-$pad);
my $parent=defined($procTgid[$i]) && $procTgid[$i]!=$procPid[$i] ? $procTgid[$i] : $procPpid[$i];
my $line=sprintf("$padL%05d%s$padR %5d ", $ppid, $procThread[$i] ? '+' : ' ', $parent);
# Handle --procopts f
if ($procOpts=~/f/)
{
$majFlt=$procMajFltTot[$i];
$minFlt=$procMinFltTot[$i];
}
else
{
$majFlt=$procMajFlt[$i]/$interval2Secs;
$minFlt=$procMinFlt[$i]/$interval2Secs;
}
my ($cmd0, $cmd1)=(defined($procCmd[$i])) ? split(/\s+/,$procCmd[$i],2) : ($procName[$i],'');
$cmd0=basename($cmd0) if $procOpts=~/r/ && $cmd0=~/^\//;
$cmd1='' if $procOpts!~/w/ || !defined($cmd1);
$cmd1=~s/\s+$// if $procOpts=~/w/;
$cmd1=substr($cmd1, 0, $cmd1Width);
if ($procOpts!~/[im]/)
{
# Note we only started fetching Tgid in V3.0.0
$line.=sprintf("%-8s %2s %1s %5s %5s %2d %s %s %s %s ",
substr($procUser[$i],0,8), $procPri[$i],
$procState[$i],
defined($procVmSize[$i]) ? cvt($procVmSize[$i],4,1,1) : 0,
defined($procVmRSS[$i]) ? cvt($procVmRSS[$i],4,1,1) : 0,
$procCPU[$i],
cvtT1($procSTime[$i]), cvtT1($procUTime[$i]),
cvtP($procSTime[$i]+$procUTime[$i]),
cvtT2($procSTimeTot[$i]+$procUTimeTot[$i]));
$line.=sprintf("%4s %4s ",
cvt($procRKB[$i]*$mult/$interval2Secs,4,0,1),
cvt($procWKB[$i]*$mult/$interval2Secs,4,0,1)) if $processIOFlag;
$line.=sprintf("%4s %4s %s %s",
cvt($majFlt), cvt($minFlt), "$padL$cmd0", $cmd1);
}
elsif ($procOpts=~/i/)
{
$line.=sprintf("%-8s %1s %s %s %s ",
substr($procUser[$i],0,8),
$procState[$i],
cvtT1($procSTime[$i]), cvtT1($procUTime[$i]),
cvtT2($procSTimeTot[$i]+$procUTimeTot[$i]));
$line.=sprintf("%5s %5s %5s %5s %5s %5s %5s %s %s",
cvt($procRKB[$i]*$mult/$interval2Secs,5,0,1),
cvt($procWKB[$i]*$mult/$interval2Secs,5,0,1),
cvt($procRKBC[$i]*$mult/$interval2Secs,5,0,1),
cvt($procWKBC[$i]*$mult/$interval2Secs,5,0,1),
cvt($procRSys[$i]/$interval2Secs,5,0,1),
cvt($procWSys[$i]/$interval2Secs,5,0,1),
cvt($procCKB[$i]*$mult/$interval2Secs,5,0,1),
"$padL$cmd0", $cmd1);
}
elsif ($procOpts=~/m/)
{
$line.=sprintf("%-8s %1s %6s %6s %6s %6s %6s %6s %6s %4s %4s %s %s",
$procUser[$i], $procState[$i],
defined($procVmSize[$i]) ? cvt($procVmSize[$i],6,1,1) : 0,
defined($procVmLck[$i]) ? cvt($procVmLck[$i],6,1,1) : 0,
defined($procVmRSS[$i]) ? cvt($procVmRSS[$i],6,1,1) : 0,
defined($procVmData[$i]) ? cvt($procVmData[$i],6,1,1) : 0,
defined($procVmStk[$i]) ? cvt($procVmStk[$i],6,1,1) : 0,
defined($procVmExe[$i]) ? cvt($procVmExe[$i],6,1,1) : 0,
defined($procVmLib[$i]) ? cvt($procVmLib[$i],6,1,1) : 0,
cvt($majFlt), cvt($minFlt), "$padL$cmd0", $cmd1);
}
$line.=$eol if $playback eq '' && $numTop;
$procCount++;
$line.="\n" if $playback ne '' || !$numTop || $procCount<$numTop;
printText($line);
$procPrinted{$ppid}=1; # string leading 0s
}
sub aggregate
{
my $pidZ=shift;
my $kidArray=$procChild{$pidZ};
foreach my $kidZ (@$kidArray)
{
my $kidI=aggregate($kidZ);
}
if ($pidZ ne '00000')
{
my $i=$procIndexes{$pidZ*1};
# Aggregate everything that makes sense...
foreach my $kidZ (@$kidArray)
{
$kidI=$procIndexes{$kidZ*1};
$procSTime[$i]+= $procSTime[$kidI];
$procUTime[$i]+= $procUTime[$kidI];
$procSTimeTot[$i]+=$procSTimeTot[$kidI];
$procUTimeTot[$i]+=$procUTimeTot[$kidI];
$procMinFlt[$i]+= $procMinFlt[$kidI];
$procMajFlt[$i]+= $procMajFlt[$kidI];
$procMinFltTot[$i]+=$procMinFltTot[$kidI];
$procMajFltTot[$i]+=$procMajFltTot[$kidI];
if ($processIOFlag)
{
$procRKB[$i]+= $procRKB[$kidI];
$procWKB[$i]+= $procWKB[$kidI];
$procRKBC[$i]+=$procRKBC[$kidI];
$procWKBC[$i]+=$procWKBC[$kidI];
$procRSys[$i]+=$procRSys[$kidI];
$procWSys[$i]+=$procWSys[$kidI];
$procCKB[$i]+= $procCKB[$kidI];
}
# If one not defined for this process, none defined and same for parent
$procVmSize[$kidI]=$procVmLck[$kidI]=$procVmRSS[$kidI]=$procVmData[$kidI]=
$procVmStk[$kidI]=$procVmExe[$kidI]=$procVmLib[$kidI]=0
if !defined($procVmSize[$kidI]);
$procVmSize[$i]=$procVmLck[$i]=$procVmRSS[$i]=$procVmData[$i]=
$procVmStk[$i]=$procVmExe[$i]=$procVmLib[$i]=0
if !defined($procVmSize[$i]);
$procVmSize[$i]+=$procVmSize[$kidI];
$procVmLck[$i]+= $procVmLck[$kidI];
$procVmRSS[$i]+= $procVmRSS[$kidI];
$procVmData[$i]+=$procVmData[$kidI];
$procVmStk[$i]+= $procVmStk[$kidI];
$procVmExe[$i]+= $procVmExe[$kidI];
$procVmLib[$i]+= $procVmLib[$kidI];
}
}
}
sub commandCheck
{
# see if user entered a command
my @ready=$proctreeSelect->can_read(0);
if (scalar(@ready))
{
my $command=;
chomp $command;
if ($command=~/^(\d+)/)
{
$pidSelect=sprintf("%05d", $1);
$pidPrinted=0;
}
elsif ($command=~/^a/)
{
$aggregateFlag=($aggregateFlag+1) % 2;
}
elsif ($command=~/^d(\d+)/)
{
$depth=$1;
}
elsif ($command=~/^([imp])/)
{
my $format=$1;
$procOpts=~s/[imp]//;
$procOpts.=$format;
}
elsif ($command=~/^h/ || $command eq '')
{
helpMenu();
}
elsif ($command=~/^k/)
{
$kFlag=($kFlag+1) % 2;
$mult=($kFlag) ? 1024 : 1;
}
elsif ($command=~/^s(\S+)/)
{
my $skip=$1;
if (defined($TopProcTypes{$skip}))
{
$topType=$skip;
}
else
{
treeError("Invalid process sorting field '$sort'");
}
}
elsif ($command=~/^t/)
{
if ($procOpts=~/t/)
{
$threadsFlag=($threadsFlag + 1) %2;
}
else
{
treeError("threads must be selected with --procopts to use 't' command")
}
}
elsif ($command=~/^w(\d+)/)
{
$cmd1Width=$1;
}
elsif ($command=~/^z/)
{
my $saveOpts=$procOpts;
$skipFlag=($skipFlag+1) % 2;
$procOpts.=($saveOpts!~/z/) ? 'z' : '';
}
elsif ($command=~/^Z(\d+)/)
{
if ($processIOFlag)
{
$ioSkip=$1;
$expSkip=$ioSkip; # for reporting name in header
}
else
{
treeError("'Z' only applies to kernels that track process I/O")
}
}
else
{
helpMenu("Invalid command: $command");
}
}
}
sub treeError
{
print "$clscr$clr";
print "$_[0]\n" if defined($_[0]);
print "Press RETURN to go back to display mode...\n";
;
}
sub helpMenu
{
print "$clscr$clr";
print "$_[0]\n" if defined($_[0]);
print "Enter a command and RETURN while in display mode:\n";
print " pid only display this pid and its children\n";
print " a toggle aggregation between 'on' and 'off'\n";
print " dxx change display hierarchy depth to xx\n";
print " i change display format to 'I/O'\n";
print " k toggle multiplication of I/O numbers by 1024 between 'on' and 'off'\n";
print " m change display format to 'memory'\n";
print " p change display format to 'process'\n";
print " h show this menu\n";
print " stype where 'type' is a valid sorting type (see --showtopopts)\n";
print " entries with 0s in those field(s) will be skipped\n";
print " wxx max width for display of command arguments\n";
print " z toggle 'skip' logic between 'on' and 'off'\n";
print " Zxx when skipping, only keep entries with I/O fields > xxKB\n";
print "Press RETURN to go back to display mode...\n";
;
}
1;
collectl-3.6.9/README 0000644 0001750 0001750 00000001315 12230241726 012355 0 ustar mjs mjs If you're real lazy, just run INSTALL and it will install collectl into the
same locations as the rpm. It will install as /usr/bin/collectl and all the
other runtime components will be placed into /usr/share/collectl. If you
really care where everything goes, read the script as it's pretty short.
There's also an UNINSTALL that will completely remove everything.
If you want to be more creative, you can either hack up the installation
script or use it as a guide to move things around to whereever you want them
keeping a couple of things in mind:
- collectl.conf is looked for first in /etc and then in its binary directory
- all ph files must be in the same directory as collectl itself OR /usr/share/collectl
collectl-3.6.9/UNINSTALL 0000755 0001750 0001750 00000001726 12230241726 013002 0 ustar mjs mjs #!/bin/sh
DESTDIR=${DESTDIR:="/"}
# R e m o v e O L D S t r u c t u r e I f T h e r e
# These structures were created in a pre-$DESTDIR world
BINDIR=/opt/hp/collectl
# This is code
rm -fr $BINDIR
# These are all symlinks
rm -f /usr/bin/collectl
rm -f /usr/sbin/collectl
rm -f /etc/collectl.conf
rm -f /etc/init.d/collectl
rm -f /usr/share/man/man1/collectl.1.gz
rm -f /etc/init.d/rc?.d/*collectl
rm -f /etc/rc.d/rc?.d/*collectl
rm -f /etc/rc?.d/*collectl # debian different
rm -f /etc/init.d/collectl # gentoo and generic
rm -f /etc/runlevels/default/collectl # gentoo
# N e w D i r e c t o r y S t r u c t u r e
BINDIR=$DESTDIR/usr/bin
DOCDIR=$DESTDIR/usr/share/doc/collectl
SHRDIR=$DESTDIR/usr/share/collectl
MANDIR=$DESTDIR/usr/share/man/man1
ETCDIR=$DESTDIR/etc
INITDIR=$ETCDIR/init.d
rm -f $BINDIR/collectl
rm -f $ETCDIR/collectl.conf
rm -f $INITDIR/collectl
rm -f $MANDIR/collectl*
rm -fr $DOCDIR
rm -fr $SHRDIR
collectl-3.6.9/RELEASE-collectl 0000644 0001750 0001750 00000340630 12230241726 014305 0 ustar mjs mjs dnRELEASE NOTES FOR COLLECTL
INSTALLATION
Installing the rpm
rpm -ihv collectl-x.y.z.noarch.rpm
Installing from source
unpack the tarball, which you've obviously done
follow the instructions in the README, which basically says to run INSTALL
Configure to start on boot
In both cases, collectl will not be configured to start on boot
but can easily be set to do so with the command:
chkconfig collectl on
KNOWN PROBLEMS/RESTRICTIONS
- There is a known problem with older perl Time::HiRes modules, newer
versions of glibc and colletcl intervals of 1 second or greater (see
http://collectl.sourceforge.net/HiResTime.html for more details) that
can result in 'setitimer' messages being logged at system startup when
collectl has been configured to run as a daemon. these messages appear
to benign, but be sure to let someone know if that proves not to be the
case. If collectl determines your system has this mismatch, it will
report it as a warning in collectl's message file in /var/log/collectl
every time it starts as a daemon. If you choose, you can easily turn
off the checking by editing the entry at the bottom of /etc/collectl.conf
named TimeHiResCheck and setting it to 0.
- if system time is changed by more then the log rolling frequency after
collectl starts, multiple log files will be created during the next polling
cycle(s)
- never include aliased networks in the network summary calculations
CHANGES
3.6.9-1 Oct 18, 2013
- typo in network plot header loop resulted in infinite loop [thanks andy]
- remove $int/secs from numa hit rate calc AND add more precision to its output [thanks stig]
3.6.8-1 Jul 20, 2013
- new flag $exportComm must be set in gexpr/ganglia so that they won't
generate an error if run without -f or -A [thanks tom]
- new switch: --intfilt allows filtering of interrupts
- always log messages of type F/E to syslog in daemon mode even if
-m is not set [thanks again, tom]
- wasn't dealing correctly with missing whitespace after network name in
/proc/dev/net in initRecord() [thanks andy]
- updated init.d script for suse per the maintainer's instructions [thanks tom]
- extra spaces were being printed in plot mode for tpc stats
- added entry to envrules.std to deal with intel Phi Co-Processor
- debian init.d script now does 'exit 1' if status reports 'not running'
- rawnetignore switch wasn't working correctly
- found/fixed some subtle problems with --procanalyze as well as some cleanup
- need to ignore first sample after initializing summary arrays
- need to init summary hashes for thrutime and accumT because get uninit var
in print routine is only a single process entry
- found a typo in procAnalyze() to a $usecs which wasn't being used!
- added error check to make sure --procanalyze with -P requires -s
- added a little more debugging output for -d128
- discovered dynamic disk/network detail names for interactive mode were not
being reported correctly. sounds a lot worse than it is because this is
typically not done very often nor are disks/networks very dynamic except in
large, virtualized environments such as clouds
- add to list of devices to exlude from network summary data: tap, dp and nl,
which are associated with openstack cinder. remember you can always add
more to that list with --netfilt
- $lastHour was never referenced and dayInit() called every time a log was
created so fix logic to update $lastHour correctly AND call initDay() one
time and do it before newLog() called.
- closed a couple of file handles that were left open and reportedly
causing some defunct processes with -sx. [thanks brian]
- fixed bug in lustre stats recording [thanks roland]
- clarified --showsubopts text about disk and network filters in that they
apply to both summary and detail data output
- fixed problem with --import and --stats
- --statsopt a didn't work because when changed some internal logic missed
changing a test of $timestampFlag to $timestampCounter[$rawPFlag] and so
now $timestampCount can be removed entirely
- clear $firstpass after 1st pass during playback
- make sure filename initialized before calling loadConfig so if there is
an error logsys() doesn't get an undefined var warning
- to be safe, remove any quotes on net/dsk filters in case included by
mistake in DaemonCommands string
- tightened up tests to see if daemonized collectl already running
- if no hiRes::Time, fudge the value of $microInterval based on -i [thanks Domi]
- new --procOpt k, removes known shells from process listing with -sZ,
currently set to /bin.sh, /usr/bin/perl, /usr/bin/python and python
- fixed varname in lexpr: $debug should have been $lexDebug
3.6.7-1 Mar 8, 2013
- set network speed for vnets to '??' so they'll use $DefNetSpeed for
bogus checks since the kernel hardcodes then to 10 which makes no sense
[thanks rick]
- code to print brief totals for -st wasn't include in a conditional
so you'd always get extra columns of output when -st was NOT included
- needed to initialize numaMem->{lock} for cases where user selects -sM
and no data collected [thanks laurence]
- added randomize [thanks robert] and align switches to graphite module
and align switch only to gexpr.ph since gexpr uses current times in messages
- added escape switch to graphite to allow one to change the dots in hostname
- change to suse startup script to look in /usr/sbin instead of /usr/bin
- added debug mask of 16 to lexpr to help test x= switch
- can now use commas OR colons with lexpr,x= though commas preferred and
colons may go away
- added disk qlen, wait, svctime and util to lexpr
- it was pointed out that in getExec() I'm initializing $oneline instead of $oneLine
[thanks joe]
- for debian init script, reverse logic for running start-stop-deamon with
-test so it will work with buxybox too [thanks chris with help from troy]
- new switch: --cpuopts z (the only option) which suppresses lines of idle
activity from detailed stats
3.6.6-2 Dec 7, 2012
- when purging imported detail plot data, only do so if file had changed
- when playing back multiple files, do NOT try to process a new file that
has not yet seen the end of the current interval ($timestampCound==1)
- fix SuSE init.d script, [thanks tom]
3.6.6-1 Nov 25, 2012
- last version broke lexpr and it wasn't correctly handling intervals
other than 1
- do not set $dskChangeFlag to 4 when maj/min numbers change as it does
not mean the stats changed
- removed checks for major/minor disk numbers changing
3.6.5-2 Sept 27, 2012
- was not updating new major/minor numbers for a disk when they changed so
got stuck in a loop which kept disk maj/min changed every interval
- new -r option to purge older .log files, def=12 months
- fixed DaemonCommands to preserver order so you can override anything by
adding on the right side of it
- new 'align' switch added to lexpr so default is NOT to align to whole min
- for -sE do not convert negative temperatures [thanks kevin]
- add error handling to 'print' in logmsg
- vmstat needs to set $sameColsFlag to make header pagination work with -p
- new graphite switch f, use fqdn for host [thanks Bryant]
3.6.5-1 Sept 10, 2012
- when lexpr called with x= it needs to set summary data flag in case
nothing else is being reported, otherwise timestamps print after the
data instead of before
- lexpr typos: $tcpError, $udpError and $icmpError should not be singular
- timestamp wasn't being updated for -sD because it was specified in $dskdetFormat
- explicitly close logs before opening new ones in the hope that the occasionally
corrputed file problems with gunzip will go away
- tcp 'last' variables weren't correctly initialized and so was printing bad data
on first line of output
3.6.4-2 August 28, 2012
- modified lexpr, gexpr and graphite such that when i= is used, to align
sending on whole minute boundaries which is particularly useful with rrd
3.6.4-1 June 25, 2012
- merged snmp and tcp stats under -st and changed export routines to
show summary error counts for -st. removed snmp.ph from kit.
summaries (based on --tcpfilt) as does brief format
- correctly deal with dynamic disks/networks
- instead of pulling names from header, get them from raw file when discovered
- simplify code that deals with changed disks, now that more cleanly handled
- replace runtime calls to 'die' with calls to syslog
- readS was still left in INSTALL! [thanks gavin]
- added system boot time to header
- new values for procopts s/S to show process start times
- graphite.ph now prints loadavgs to 2 decimal places [thanks brandon]
- extended lexpr,x= functionality to also call an init routine
- initFormat now returns entire header!
- if nothing returned from an import module on a printVerbose or printPlot call
for detail data do not call printText() since it will screw up colmux and
plot detail file with empty lines
- new --rawdskignore AND --rawnetignore because sometimes easier to specify
a pattern of things to ignore
- removed restriction for running as root to get network speeds via ethtool
by looking in /sys/devices now
- slight change to way the disk queue depth is being calculated to provide
better accuracy [thanks ken]
- new --dskopts f reports disk details with some fractional values
- always calculate disk details even when only doing -sd since a plugin
might want to get at them
- new graphite switch b, will cause output to be prefaced by a specified string
[thanks justin]
- slight change to s= functionality for lexpr, gexpr and graphite: no arguments will
disable all but imported data, allowing you do log -s data to files sending over socket
- need to give other routines (specifically --import) access to the lexpr
interval by declaring it with 'our'
- had to change the way lexpr/gexpr/graphite do min/max/avg since they were
using a positional index to track intermediate values when clearly a hash
is required for cases where not all intervals contain same elements
- -P and --plotflag had different effects on $headerRepeat because prior to calling getopts
I was peeking ahead for an ARG of -P and not including --plo [thanks devilized]
- gexpr module has wrong units for network packets and with 'g' modes had to multiply
kb counts by 1024 to convert to bytes, which is the units for these that ganglia uses
[thanks, trevor]
- clean up handling of missing ipmitool and root access [thanks trevor]
3.6.3-2 May 01, 2012
- finally remembered to remove readS from the kit [thanks joseba]
- when filtering a process by the fill path with 'f', never include collectl itself
- documented utime in manpage
- if -i0 set $DefNetSpeed to 0 so we don't throw any 'bogus' network speed messages
- new switches, --rawdiskfilt and --rawnetfilt, allow one to filter disks/nets at time
of data collection so they never appear in raw file
- added call to IntervalEnd() (if it exists) for --import
- add option timeout to --address when connecting back to explicit address
- moved code that deal with fractional intervals and !HiRes closer to other interval processing
- added 'strict' to snmp module as well as 'help' option: snmp,h
- fixed problems with --import
- if --import is used to generate detail data with -f and -P not specified, collectl throws an error
trying to close the detail log which clearly hasn't been created
- when using interval other than the defaul AND -s-all, blank lines are printed for standard intervals
which don't have imported data. this applied to brief, verbose AND detail data
- added some more systems to envrules: Proliant SL230/SL250 Gen 8 and SE1170s
3.6.3-1 Mar 03, 2012
- fixed serious bug introduced a number of versions ago, which during playback of multiple files
and specifying date/time caused collectl to continue reading first timestamp in each file
and generating 'uninit variable' errors. not harmful, but inefficient and ugly!
- added exit codes of 0/1 to all the exit points
- moved help text for --stats from basic to extended
- found $file=~/rawp/ near line 1440 clearing $1, $2 and $3 and so $prefix, $fileDate and $fileTime
were not getting set correctly
- clarified 'No files processed' message to be a little more explicit
- broaden where collectl looks for lustre modules and also fixed a typo
of $lustops to $lustOpts [thanks brian]
- procAnalize incorrectly totaling fault totals instead if interval values [thanks andy]
- limit sizes of -procfilt for username/command to 19 and 15 respectively
- change order of ps command in loadPids() so they return max length fields for user/command
- remove () from command field from /proc/pid/stat in pidNew()
- optimize new pid processing with --procfilt
- add new pids to pidSkip{} as appropriate
- undef pidSkip{} whenever pids wrap
- added hello.ph and graphite.ph to INSTALL
- was incorrectly setting DiskFilterFlag to 1 all the time, even when not overridden in
collectl.conf. while not a bug, it does cause a slight increase in overhead
3.6.2-1 Feb 28, 2012
- changed behavior of --runas to no longer require a change to /etc/init.d/collectl
as it now uses /var/run to write collectl.pid into. this means to ineract with a
non-root daemon, you still need to be root, which makes sense.
3.6.1-4 Feb 20, 2012
- removed --ssh switch, making detecting the parent going away the default behavior
- added switch --nohup which will allows collectl to continue running if parent exits,
which is more consistent with how --nohup itself works
- in logmsg ONLY write to STDERR when attached to a terminal
- serious problem when using --tworaw and a flush interval < that for the process data
occurs because newer versions of zlib will fail if you try to flush to a file that
has not been updated. since I don't know which version of zlib this started happening
in and feel this is a relatively rare case, we're just rejecting this combination
regardless of zlib version. I do have an email out to the zlib author and if I ever
get to the bottom of this will be ble to relax this restriction.
- use getimeofday() for timestamps in logmsg()
- enhanced timing parameters when -i0 used. if specified user 2nd/3rd parameter as ratio
to first making it possibily to measure loads of different rations other than 1:6:30.
- discovered --import was missing from man pages and so added it
- when playing back a file, set $verboseFlag if user specified --verbose but NEVER clear it
- experimental import: snmp, see http://collectl.sourceforge.net/Snmp.html for details
- printf in record() blows up if formatting chars in command string! [thanks mike]
- added accumulated time as a --top sort option
- changed formatting of accumulated time in process output to simply be hh:mm:ss or
mm::ss.ss when less than an hour to be more in line with top
- new swithes, --stats and --sumstats report stats in brief mode, the latter only summary
data
- during playback need to check $numProcessed before reporting none were processed
- stats reporting logic wasn't processing 1st file, checking for $numProcessed>1
- removed -oA and replaced/extended functionality with --stats/--statopts
- wasn't allowing --procopts playing back process data unless -sZ which was silly
- subtle problem found: illegal 'last' in pidNew() because file disappeared between initial -e
and trying to open it a few usecs later! can't exit a sub via last so changed to return(0)
- our friends at OFED slightly changed the output of perfquery again [thanks frederic]
3.6.1-3 Jan 13, 2012
- added 'Reason: $!' to socket open failures
- was not reporting interrupts in playback mode correctly
- added $memAnon to lexpr
- need to initialize $thisConfig when --lustopts set [thanks joe]
- do not allow -f with gexpr and not one or both of -P/--rawtoo [thanks again, joe]
- modify misc.ph to honor --showcolheader
- modify lexpr, sexpr, gexpr to reject --showcolheader
- if --showcolheader and --export (only works with vmstat for now), exit after first
print call
- remove restriction of not letting someone use --home with proc/slab data since they
may want to apply filters and therefore not need more than a terminal full
- new switch, --comment, allows a user to add a comment to the header
- only read /proc/slabinfo IF slab monitoring requested AND if slab monitoring requested
make sure /proc/slabinfo is readable (some admins only allow root access)
- added code for slow proc read speed test on all system >= 32 CPUs except for RHEL6.2
and SLES 11 SP1
- if /sys/devices/system/node doesn't exist, set CpuNodes to 1 and disable -sM if set
- fixed a lot of typos in a lot of docs
- only set a socket failure handler with a socket is explicitly being opened
- added 'h' option to gexpr, lexpr and sexpr
- changed the way vmstat.ph decides to print its header
- added new process option: x, which adds extended data to standard display
- added Mlocked to verbose memory output as well as numa stats [for fred]
- changed root name for cpu detail data in gexpr from cputotals to cpuinfo [thanks evan]
- new export: graphite
- normalization for CPU load reports jiffes instead of a percentage [thanks guy]
- removed restriction against using -D as non-root user
- as per https://bugzilla.redhat.com/show_bug.cgi?id=716825, non-root access to /proc/pid/io
is now considered a security hole and so may not have read access! therefore we need to
check to see is the io structure is readable before trying. if it isn't, zeros will be
reported for non-readable structures
- new procopts option I, disables collection of IOSTATS and reading of /proc/pid/io,
a performance optimization at the expense of less process information
- newswitch, --runas will cause collectl to run as a non-root daemon. this WILL require
changes to the init.d script to work! be sure and read the man page
- changed location where $doneFlag was getting cleared because stopping the daemon before
initializtion was completed was causing the flag to be reset to 0 and not left at 1
- change sort limit for process counters from 6-9s to 9-9s [thanks stig]
- added SUSE SP info to header
- added debian and ubuntu release/distro info to header
3.6.0-3 Oct 17, 2011
- added dirty memory to lexpr
3.6.0-2
- support for numa
- split anon pages into separate field in verbose mode as well as plot format
- changed the memory header for -sm to SUMMARY rather than STATISTICS as the
latter is currently used to indicate detail data, something that didn't exist
for memory prior to numa support
- added --xopts i to be consistent with --dskopts and --netopts. did NOT add
such a switch for lustre
- expanded error checking with perfquery to catch 'Failed to open' errors
during initialization
- discovered and removed reading of /proc/stat during -sm, which was there to support
2.4 kernel fields that have since been moved
- changed collectl-debian start script to use /bin/sh instead of bash
- removed ".B collectl" at start of collectl man page for debian/lintian compliance
- made width of number of dentries in -si --verbose 7 instead of 6 digits wide
3.6.0-1
- do NOT call derived() when playing back rawp files or you'll get unit var
for $memUsedLast.
- need to include non-numeric type interrupt counts in interrupt totals
- fixed a few problems with envronmental data and interpretation of --envopts
- was not allowed to use with -P and only 'M' should have been restricted
- was only honoring C/F when temp name started with Temp rather anywhere in string
- was not correctly overriding default ipmi devices with user define options
- fixed formatting/calculations for interactive memory subtotals generations when
RETURN is typed in conjunction with --memopts R in brief mode
- added new section to FAQ called 'gottchas' as a place to describe the perils of
round-off error and normalization
- when printing verbose data in import modules, need to clear $$lineref or the last
line that mainline collectl reports (if any) will be repeated. this was fixed
in hello.ph and atigpu.ph
- new switch: --dskopts z, which when specified filters out disk details lines of all 0s
- added switch examples to start scripts for clarification of use
- added support for 'vd' disks [thanks gavin]
- since kernel 2.6 compatible with 3.0 and 2.4 is sooo old, 2.4 support officially dropped!
[thanks for the push, tony]
- dropped support for collectl data generated by versions of collectl older than 2.0
- need to set $cpusEnabled to 0 when playing back interrupts in plot format w/o -sC, since the
code that normally does that has already been executed and 'C' not yet added to $subsys. subtle...
- filled in some missing ; in nvidia.ph in PrintPlot routine
- fixed problem writing plot files with --import
- added 'i' to both dskopts and netopts which will cause i/o sizes to be displayed in
brief mode like --iosize except in this case independent of each other
- do not include virtual networks in network summary [thanks hank]
- in newLog() need to use gettimeofday for current time when hires::time is used otherwise you'll
occasionally get a time 1 second earlier and new files names are wrong! [thanks hank]
- exclude vlan from network totals to avoid duplicate counts [thanks andrey]
- added 2 new fields to verbose cpu Summary Stats - Run Total and Blocked Total
- added VmSwap to process/memory display
3.5.1-1 May 23, 2011
- change expression used to find CPU count in /sys since -P isn't necessarily
built into all greps
- instead of only getting the platform name when -sE, always try to get it
- forgot to include 'T' as valid --envopts
- check for failure of 'ipmitool sdr dump' command
- need to ignore interval checks with --showcolhead and -sE
- fix bug in checkSubSys() because while it could find newer subsys it couldn't find
dropped ones
- needed to clear nethostflag outside conditional that looks at prefix changed, which
was incorrectly preventing consecutive files on the same day from being identified
- added new routine pushmsg() that allowed one to stack up messages generated BEFORE
'beginning execution' message and then play them afterward, making log easier to read
- changed several calls from logmsg() to pushmsg()
- added support for files that cross midnight and ability to play them back in full
see updated Playback.html
- remove duplicate message in sexpr
- have found an instance where the number of networks in the header didn't match the ones
listed (some were dropped!) and so added a check to take care of this
- renamed $active, $inactive and $dirty to $memAct, $memInact and $memDirty for better
consistency with other memory variable names. Didn't bother with older V2.4 mem variables
- new switch --memopts R: display memory info as changes/interval, similar to sar's -R switch
- logic to clear '$sameColsFlag' in verbose mode and --import was wrong
- --showcolheaders and -sE requires root
- added support for nvidia driver V270.41.19 which has different output format. highly
probable other versions will behave different as well
3.5.0-3 Feb 12, 2011
- expanded interrupt details to include non-numeric interrupts
- new import module added for GPUs: nvidia.ph
- added getExec type 0 to support new import
- updated version of gexpr, with new switches to control using default ganglia
variable names
- bug fix: wasn't sending E and F types messages to syslog
- wasn't initializing enough 'last' vars for latest nfs V4
- only allow -sT with -P or -f
- added new switch --tworaw as a synonym for --group which makes more sense
- if an imported module returned -1 in its init routine, disable it. return
1 for success
- new --procopt: R causes real-time priorities to be displayed rather than RT,
at the cost of 2 extra columns in the display [thanks lee]
- added optional callback GetHeader to --import API, if not defined not called
- change error handling when playing back files with no selected subsystems to be
non-fatal, skip the file and continue processing
- added dl585-g7 to envrules.txt
- allow -s-all to remove ALL L subsystems when you only wanted --import data played
back. I actually forgot to add this to release notes until V3.5.1
3.5.0-2 Jan 09, 2011
- turned utime into a mask, so we can control the granularity of micro-logging
to include /proc time with/without process accesses
3.5.0-1 Jan 09, 2011
- renamed --showplotheaders to --showcolheader since it now applies to ALL headers
for single header line output (will only show cpu for -scd --verbose)
- fixed ALL verbose and detail output formats to include date/time headers
- newer kernels added additional files to /sys/devices/system/cpu/ which messed
up the way total CPUs were being calculated
- added 2 new variableS to lexpr: cputotals.num and cputotals.total [thanks chris]
- removed unused switch --pidfile from collectl -x
- file processing push/pop code wasn't handling data change correctly
- added new flag to show host changed since THAT was what was needed in 'consecutive'
file identification processing
- found problem with playing back multiple files with --thru for different hosts!
needed to 'undef $newSeconds[$rawPFlag]' whenever hostname changed
- new netopts values
e - show errors in brief mode and explicit types everywhere else
E - only print lines that have non-zero network errors in them
- new diagnistic switch --utime, causes periodic micro-timestamps to be written into
raw file at different points in time for finer grained measurements of operation times
3.4.4-3 Dec 9, 2010
- if -s during playback, at least ONE requested subsys must be in recorded file.
if c recorded, C would cause error message because pattern match didn't have 'i'
- add requirement for STDOUT to be connected to a terminal as a condition to call resize
- change to collectl.conf - roll logs at exactly midnight, not 1 minute past
- new --envopts value of T to truncate values to integers
- ignore 'Fan Redundant' in env data for dl160g6
- if impi data field is blank, ignore it
- fixed filtering of ipmi data AND renames 'c' option to 'p', for power
- include THRD in -P format for processes
- only turn off echo when in brief mode AND not playing back a file
- if data collectl w/o HIRES and display request msec, set default to '000' instead of 0
- discovered only --ssh in help so removed -S
3.4.4-2 Nov 10, 2010
- base36() needs to do an int() on values <10 so their fraction not
included in output string
- reduced printing of headers for -sf --verbose to one call to printText()
per line. otherwise one hostname prepended to each line of socket call.
- fixed a problem with --procfilt C: it was trying to match whole process
name rather than just the beginning of it [thanks gary]
3.4.4-1 Nov 09, 2010
- vmstat not handling date/time correctly, needed $dateTime[0]
- need to call export module's init routine in playback mode
- lustre 1.8.4 module location moved, check expanded [thanks Frederik]
- new top sort options, pid and cpu, which don't make a lot of sense
unless used with filters
- do NOT include hostname in RECORD printing routine with -A
- CPU verbose output should not right shift 1st header line with -oT
- removed printing of extra '$line' at end of NFS DETAIL header
- incorrectly setting recSubsys to [YZ] if user specifies --top even
if -s specified too! They should be merged [thanks mats]
- don't write to a socket if shutting down in which case $doneFlag set
- don't report socket errors if not in server mode
- added 'ProLiant DL160se G6' to envrules.std
- disableSubsys should ONLY remove subsystems from export option 's='
was also clearing KFlag rather than LFlag [thanks chris]
- new process sort option 'thread', sorts by thread count
- changed start/stop in initd scripts from "$network +openibd" to "$all"
so collectl will start after everyting else
3.4.3-3 August 19, 2010
- added --netfilt
- very rare: if playing back CPU data but none collected, be sure to
set $cpusEnabled to number of CPUs or else you'll get warning that
one or more disabled
- pattern match wrong for 'emcpower' disks [thanks lewis]
- changed disk details to use 'cvt()' for reporting number of I/Os since DM numbers
can be more than 4 digits
- change --umask behavior. default is to do nothing unless explicity set
AND user is 'root'
- 2 new process sort fields: pid and cpu
3.4.3-2 August 16, 2010
- only look at $cpuDisabledFlag when processing CPU data
- perfquery in OFED 1.5 can report warnings in its output stream which need to be ignored
- if you try to playback a file and specify -s with no existing subsystems you'll
get an error
3.4.3-1 August 02, 2010
- perfquery checks problems
- version finding code not working correctly for ofed 1.5
- disabling -sl by mistake when perfquery not found
- when errors detected during initialization not skipping subsequent checks
3.4.2-5 July 21, 2010
- changed INSTALL to only execute commands like chkconfig OR update-rc
when $DESTDIR is / [thanks mike]
3.4.2-4 July 09, 2010
- added --dskfilt
- added check for client-side OST uuid status 'DEACTIVATED', which seems to
have showed up somewhere in the 1.6 timeframe but now sure when, thanks Heiko
3.4.2-3 June 25, 2010
- new memory field 'SUnreclaim' ONLY available in plot format and lexpr,
just not enough room in terminal based output [thanks seb/fred]
- misc now considers uptime, mhz and mounts as 'lightweight' counters and will
sample every standard interval. Only logins, which is heavy-weight, will be
sampled based on "i=" or the default of 60 seconds. Further, all lightweight
samples will be returned every interval by lexpr whereas the heavy-weight ones
will only be returned when sampled. In order to keep sexpr/gexpr formats constant
(primarily because I don't know the effect of not doing so), they will report
all counters every interval.
- support for CPUs dynamically changing stats and going off/on-line
- NOTE -- can't detect this during interrupt processing unless also
monitoring CPU data, which people typically do anyways
3.4.2-2 June 15,2010
- not correctly handling discovery of new disks during playback
- new feature: select process by UID range [thanks mark]
- fixed bug in --procfile u/U processing while testing
- added systot and usertot to lexpr to report totals for all system and user
counters
- changed error message processing when trying to playback a file with process
when there isn't any or slabs data, etc. Rather than only show the message
when -m, which could result in only a 'no files processed' message they will
be unconditionally displayed as they should
3.4.2-1 May 21, 2010
- change default umask to 133 so that colplot can read files since webserver
doesn't have privs
- now that raw files are always compressed, the message about disabling it
with -oz when no compression no longer makes sense so the message has been
clarified to use --quiet with raw files and -oz with plot files
- added README-WINDOWS to src tarball
- cleaned up code that still expected [com] in $lustOpts instead of $lustreSvcs
- more cleanup and bug fixes to INSTALL for debian support. thanks bernd
- change to /bin/sh
- do not use ANY explicit paths
- minor changes to man pages, also for debian restrictions
- wasn't reading NfsFilter correctly from header on playback
- save perfquery version and use it to drive the skipping of 'field 13' rather than
OFED versions which isn't always available
- do not issue 'stty' if !PC, running on terminal and !background. missed a couple...
3.4.1-5 Mar 30, 2010
- new env options F/T converts temps to C or F
3.4.1-4 Mar 29, 2010
- new switch --whatsnew prints a summary of changes, a mini-release notes
3.4.1-3 Mar 23, 2010
- added Fusion-IO card to list of valid disks: fio
- gexpr, lexpr and misc weren't honoring internal interval counter.
- if a secondary/tertiary interval specified gexpr/lexpr didn't process
it correctly
- new switch: --envfilt allows you to specify filters
- if you specify a " in DaemonCommands it gets passed along in the variable itself
(not a problem for ') so we have to remove them
- added new section 'Filters' to header. Added EnvFilt and moved NfsFilt to it
- added new switch --envremap, which allows for renaming one or more output field names
- added new feature switch to lexpr. if x=file is specified, that file will be loaded
via require and a corresponding function name called after every print cycle, allowing
one to do modified, custom output
- new switch, --umask too control output file protections, see man umask. default is 0137
- new environmental option - if you include a device number with --envopts use THAT as a
device number with -d when running ipmitool. for some systems the default devices is
the slower one and this will have an impact on how fast ipmitool will run, possibly
slowing down collectl
- added 'use 5.008000', which should have probably been there years ago
3.4.1-2 Mar 16, 2010
- do now allow -oA in verbose mode
- consolidated all code to disable -s subsystems when a conflict consolidated into
disableSubsys which ALSO disables them in s= clause of --export
- removed code to disable s= in all the ph export modules since now redundant
- support for DESTDIR env variable in INSTALL/UNINSTALL [thanks Bernd]
- Voltaire changes output of ofed_info so we have to process IB version
slightly differently
- change lustre message about needing -L to --lustsvc
- changes to lexpr to include processes in run queue and to change prefix
for proc creates/runs to 'proc'
- changes fo misc.ph to ALWAYS report latest values in --export as well if 'a'
paremeter, noting the default is to only report them when sampled. collection
still defaults to 1 minute, overridable via 'i='.
- since loading formatit.ph moved in a recent release, any calls to error()
before it's loaded since it needs a routine internal to formatit. so now
only call printText() from error() if formatit loaded.
3.4.1-1 Feb 22, 2010
- when printing plot data to files, wasn't putting headers on subsequent days' files
3.4.1-0 Jan 10, 2010
- make sure all major release settings in RELEASE-collectl have dates
- remove blank line in all collectl start scripts right before 'END INIT INFO'
since debian doesn't like it and we should be consistent
3.4.0-4 Jan 04, 2010
- updated envrules to include additional parsing rules for dl185 [thanks evan]
- changed envrules header for dl585 G1 to G5
- if running an ofed >= 1.5, ignore 'CounterSelect2' field, which is right in the middle
- send errors in getExec() to /dev/null because perfquery for > ofed 1.4 is braindead
- was incorrectly using 256 to print IB debugging info instead of 2
3.4.0-3 Dec 14, 2009
- was not clearing right variable for CPU Detail Totals in sexpr.ph
- fixed typo on QLogic HCA name from qlib to qib
3.4.0-2 Dec 13, 2009
- fixed typo of HugePages from HughPages [thanks Frederic]
- fixed typo of 'openib' in start script LSB headers to 'openibd'
- clarified help and man page for --all to indicate ONLY summary data will
be reported, meaning NO process or detail data either
3.4.0-1
- restructure installation directories to be more standard
- pid was not properly set for suse flush command
3.3.7-1 Nov 26, 2009
- added support for psv [polyserve] disks
- added support for QLogic IB HCA
- changes to INSTALL/UNINSTALL to handle gentoo and to restructure 'generic'
distro processing for more flexibility in the future
- 3 'standard' tools turned out not to be standard on gentoo and so:
- limit checking for ethtool to writing to log file OR --showhead
- if can't find lspci during -sx processing (and -sx IS a daemon default),
disable -sx rather than throw a hard error.
- only use dmidecode if -sE and if not found, set product name to 'Unknown'
- creating /var/log/collectl in INSTALL so when installed this way the
daemon writes logs into that directory instead of /var/log. this now
matches what an RPM install does
- if required include files can't be find in same directory as collectl, look
in ReqDir which is initially set to /usr/share/collectl. This can be
changed in collectl.conf
- when exiting due to a fatal error, be sure to exit(1) and not just exit.
- some process I/O counters found to be missing on CentOS 4.8 and so had to
initialize to 0 in case not found
- wasn't catching 'ioall' as invalid --top option
3.3.6-2 Sep 16, 2009
- if printing interrupts in brief mode, Cpu headers have to be changed as the number
of cpus increase to 2 or 3 digits. [thanks Aron]
3.3.6-1 Aug 19, 2009
- changed error message about missing ethtool or lspci to just ethtool since
missing lspci was already caught and reported
- change location of collectl to /usr/bin in collectl-debian
- make -P honor --hr which it currently does not [thanks giles]
3.3.5-4 Jul 20. 2009
- performance optimizations in dataAnalyze()
- check process/slabs first whenever type is proc/slab. then in a separate clause
look at subsys, thereby preventing parsing of type in other checks
- always include test of subsys and do it first. found to be completely missing
in lustre tests
3.3.5-3 Jul 17, 2009
- expanded meaning of -G to include slabs in 'rawp' files and to add 'g' to the Flags
in the header, which also uncovered a number of bugs in the way batches of files for
different hosts/dates were selected/handled even before slabs were added
- drop support for -sy in brief mode since it really doesn't make much sense and if you
do specify -sy it now forces verbose mode. see Slab documentation for more on playing
back files generated with -G
- if can't find an ofed utility AND rpm isn't on system, don't use it [thanks seb]
- fixed some problems with -oA processing
- removed a couple of error checks for switches that don't apply to a particular option
since they are silently ignored already, making it easier to recall a command and add
switches rather than having to remove those that don't apply
- flush STDIN at startup in case someone typed extra CRs
- added col2tlviz to kit
- changes to --export processing broke --vmstat so moved call to setFlags() from right
before playback code (which sets them itself) to right after call to $expName init routine
- changed start scripts so that if you can specifice "start/restart {[extension] switches]"
making easier to use/document. the old syntax which put the switches 1st meant you had to
use "" if you didn't want to change them AND it didn't work with redhat's 'service' command
3.3.5-2 June 30, 2009
- added client.pl to examples/ and moved readS to /examples
- added new switch --procstate, which allows you to limit process displays to
only show those processes in one or more explicit states
- incorrectly looking for 'LustreVersion' in header instead of 'CfsVersion'
- when dropped SubOpts from header it broke pattern matching for subsys in
header during playback
- only calculate disk detail stats using CPU time when hires not available
- when reporting a lustre server that is both an MDS and OST in brief mode,
the 2nd line column headers are reversed for the types of server
- removed obsolete switches (and warnings) -b, -e, -oP, -Y, -Z, -O, --subopts and -sLL
- changed buddyinfo headers in verbose, plot and detail files being sure to include
name/zone after : in details [thanks bayard]
- use mergeSubsys() everywhere $userSubsys is used to reset value of $subsys
- changed some instaces local variable $file to begin sorting out of local variables
with the same name as the global one
- if newlog starts and NOT an interval 2 interval, we don't record correct slab data so
only clear $newRawSlabFlag (also renamed for clarification) during interval 2
3.3.5-1 June 19, 2009
- print load averages to 2 decimal places in plot format to match interactive format,
which also required adding to lexpr and allowing it to deal with fractions
[thanks stevef]
- when disk order changes, error message was not reporting correct old maj/min
numbers [thanks philippe]
- code for including >ignore< stanza in envrules was causing unititialized variable errors
- do not make sure ipmi available when running with --envtest
- do not include ':' in lexpr network name string
- re-enable sending startup and E/F messages to syslog
3.3.4-5 June 14, 2009
- old redhat distros don't recognize the -p switch on the start script so check
first before using it
3.3.4-4
- make sure all LSB headers the same and only contain "$network +openib" for
services so that collectl can run diskless and not require ntp
3.3.4-3
- fixed a few things with gexpr.ph
- incorrectly used ' instead of " for detail counters variable names [thanks evan]
- using wrong variable name for interrupt totals by CPU
- changed way lustre OST names are parsed so that they handle embedded _s correctly
- include LSB comments in start script headers
- make SubsysCore in collectl.conf match real subsys core, even though just a comment
3.3.4-2
- changed all hardcoded occurances of /etc/collectl.conf to $configFile even in
error messages, in case someone ran with -C [thanks philippe, for this and others]
- added DiskMaxValue to collectl.conf, with default of -1. If >0 and a disk
read/write rate is greater, reset all stats for this disk to 0 because something
reset them and they're probably all bogus
- moved code that initialized disk names to separate subroutine and added logic to
save disk major/minor numbers so it can also be called later if disks are reordered
- if DiskFilter specified in collectl.conf, use that string for disk filtering. if
not specified continue to use separate if statements for tests in getProc() since
they're slightly more efficient
- if diskremap.ph exists, call internal remapDisk() routine when disk array is being
initialized in initDisk()
- newLog() was clearing $printHeaders instead of $headersPrinted
- if playing back multiple files for same day with -sD and disk config changes, generate
an error if not -ou because mixing the data in the same detail file will make it impossible
to interpret
- remove unused variable '$intFlag'
3.3.4-1
- added "ProLiant BL490c G6" to envrules as a 'standard' system since there
is nothing special to do to parse the data
- changed lustreMDS data for sexpr, lexpr and gexpr to be consistent with what
is being reported. this wasn't done when lustre 1.6 support was added and
should have been
- fixed a typo in a lustre ost variable name in gexpr
- don't just report ETH traffic in -sn brief mode, use same numbers as --verbose
- added [ignore] stanza to envrules to allow ingoring anything that matches
- only call loadEnvRule is -sE or debugging with --envtest
- rewrote formatting code for g/G option because it wasn't working correctly for
all situations
3.3.3-1 April 28, 2009
- forgot to include misc.ph in INSTALL
3.3.2.1 April 28, 2009
- screwed up $rootFlag and set to 0 after it was intialized correctly
- fixed a couple of problems in INSTALL: added 'q' to gzip, added gexpr/envrules.std
- added DL385G5 top envrules.std
3.3.1-10 April 27, 2009
- If root, add product name from 'dmidecode' to header
- If !root, don't allow -sE because ipmitool will fail
- When running -sE and no --envrules, look in 'envrules.std' for matching product rules
- remove '.' from ipmi device names before applying parsing rules (screws up =~//)
- change ipmi value of 'no reading' to -1
3.3.1-9 April 24, 2009
- When splitting off the daemon options, needed to include ',2' in the split
or any *expr options get screwed up since they can have their own =
- removed 'C' from -s in daemon command string since no longer needed
3.3.1-8 April 22, 2009
- renamed cmuextras to misc and renamed all variables accordingly
- added inactive memory to lexpr
- set default interval for 'misc.ph' to 60 seconds
- a couple sets of data names in gexpr (for cpu and disk detail) were framed in
single quotes and neede to use doubles
- wrong variable name for $intrptTot
- removed check for CPU data in presence of -sD since always there
- -sL --lustopts O not properly parsing read/write bytes for CFS/SUN release
- accidentally left some debugging code in that changed 'sd' disks to 'xvd' disks
- added support for disk types of 'emcpower'
- when running with -P and --rawtoo, collectl only write to the raw file but still created an
empty prc file. Not it doesn't create that empty file. Also added reason to FAQ
3.3.1-7
- removed memhuge from cmuextras and added to core memory stats as well as
gexpr, lexpr and sexpr
- cleaned up a couple bugs in gexpr for i= processing
- silently remove 'x' from 's=' in gexpr, lexpr and sexpr if not part of -s since it
could have been disabled. this allows one to specify -sx as well as s=x without
fear of getting a hard error from the *expr
3.3.1-6
- updated collectl-debian
- added avg/min/max options to gexpr and lexpr
- added import 'cmuextras.ph' to kit
- removed line that set $message to 'unexpected perfquey error' which was
clearly the wrong thing to be doing
- in 3.2.1-6 added 'unexpected message' for perfquery failures that was wrong so
removed it
3.3.1-5 Apr 09, 2009
- need to include command switches when changing process name
- rewrite of all the start scripts (collectl, -generic, -debian and -suse) to support
multiple daemons. In the process fixed a bug where debian wouldn't restart correctly.
Added --restry 2 to start-stop-daemon and that seemed to fix it.
- added type 4 to gedtExec()
3.3.1-4 Apr 06, 2009
- changed interface to sexpr and lexpr to more closely reflect gexpr dir/file naming,
updated documentation and also changed lexpr to include only sending changes and
handling TTL, mainly by stealing a lot of code from gexpr.
- got rid of --expdir since that now handled with 'f=' option to all 3
- Had to move calling of ${export}Init to after initRecord()
- Reporting incorrect variables for -si with all 'expr' routines. Had changed
inode data a long time ago but apparently nobody uses 'expr' or -si or both
- Needed to add -sC with -sj in sexpr
- Added SwapFree to *expr even though it can be derived
- new switch: --pname name, tells collectl to run as a different process name
and use a different pid file with that name, which in conjunction with hacking
up another init.d/collectl file will allow you to run a second instance of a
daemon with a different name
- reset $interval2SecsReal to 1 at same as $intereval2Secs when $i2Secs is 0
3.3.1-1
- when writing to plot files not including new headers on subsequent days
- typo on major fault display string in lexpr.ph
- if only logging plot detail data, was getting errors trying to print to
unopened tab file
- API for --import allows custom data collection, includes example hello.ph
- had to allow for playing of file with blank Subsys field
3.2.1-6 March 03, 2009
- added --nfsopts z to filter lines of 0 in -sF mode
- if collectl.conf is not writeable (eg in R/O filesystem), do not try to add
IB paths dynamically
- wrong logic for handling --nfsopts z
- minor formatting changes to column positions in brief format and slab detail
- wasn't including CPU type, speed, cores and siblings when converting to plot files
- dropped inode info from header which was dropped from collectl awhile back
- don't report open failures on nfs data since not always there
- add support for XEN xvd disk types [thanks brian]
3.2.1-5
- incremented $nfsCommit instead of $nfsCommitTot
- wasn't handling --nfsfilt correctly on playback of 3.2.1-4 files
- don't set $sockFlag until after socket opened otherwise we can't report socket errors
on terminal
- if read & write fields for an nfs version are both zero assume not active and don't
report in detail format
- make nfs one of the default subsystems to collect data for
- UNINSTALL wasn't removing link to start script on Debian
- file selection logic for playback wasn't working correctly for multiple hosts with
multiple files on same date
- fixed preprocessPlayback() to deal with +/- when -s specified
- fixed very subtle bug involving playing back multiple files for same day, the first having
-sy and the second having -sY and -s overrided with -s+. caused print on opened filehandle
3.2.1-4
- always write client/server nfs data, using nfsc- and nfss- as prefix
- added --nfsfilt to control details output
- other misc stuff for support of ALL nfs data in raw file at once
- dropped SubOpts and NfsOpts from header
- added NfsFilt to header
3.2.1-3
- do now allow -O any more, must use --nfsopts and --lustopts
- support for nfs V4. will now collect ALL data in /proc but still only report
on 1 type either interactively or during playback, based on --nfsopts
- only turn echo back on in error() if not a PC
- only look for passwd file when recording/playing back process data
- when playing back a file with a prefix in front of the host name and specifying
multiple directories the destination was not being correctly resolved.
3.2.1-2
- only set $nfsOpts from header during playback if -s wasnt' specified OR it
was and contained an 'f'
- do not exit on broken pipe if "-A server"
- --vmstat wasn't respecting --hr 0 or 1
3.2.1-1
- fixed a couple of bugs in INSTALL
- init.d scripts and release notes copied to wrong directory
- added Passwd to collectl.conf which if defined will point to default passwd file
- changed the way /proc/vmstat read to get more data
- added swap in/out and page faults to verbose memory display
- added page faults to tab file
- when running interactively over multiple days with -P, headers were not being
including in subsequent files
- changes some verbose summary headers to mix-cased
3.1.3-1 January 23, 2009
- output for '--procopts i' off by one column near accutim
- if RETURN entered in brief mode before 1st interval reported, ignore
it because we'll get a divide by 0 error
- add +openibd to sles startup script so collectl will start after IB
- fixed problem processing data from different time zones with new
--from/-thru processing
- fatal bug in playing back process data was missed before release
- another fatal bug in --procanalyze. if looking at a process which were only
there for a single interval, when calculating the % of cpu which takes into
account the process lifetime (in this case 0), you get a divide by 0 error!
the fix is to set the duration to 1.
- not all files were opened if -s specified with + and --procanal/--slabinfo
so added restriction against doing so
- when playing back interrupt data in plot format you have to include -sC and this
was too confusing so just silently (unless -m) adding it in and documenting in FAQ.
- if --slabanal or --procanal but no -sY/Z, don't write to slb or prc file
- allow --passwd for ALL situations since /etc/passwd not valid for NIS. also add
to help output
- selection of task by UID wasn't working
- if uid can't be translated to a username, report the UID instead of ???
- fixed problem with divide by 0 errors if proc/slab analysis on multiple host/days
3.1.2-4 January 20, 2009
- bug fixes to handling of interval times
- -sm --verbose needs 1 extra line with --top
- when exiting from --top, move cursor to bottom of display
- if playing back files for same host, don't reset header counters between them
- ignore parent process when looking for duplicate instances of -sx [thanks kaya]
3.1.2-3
- support for allowing multiple clients to connect when in server mode
- new documentation page: Genenerating Plottable Files
- dropped support for data files generated by pre V1.3 version
- when rolling logs, write a timestamp onto end of last file
- in playback mode, if last timestamp of previous file matches first timestamp of
new file, treat as contiguous data which results in no 'holes' in output stream
3.1.2-2
- check for nfsopts/playback in checkSubsysOpts was incorrectly looking at
$plotFlag when it should have been looking at $playback
- added Power Meter (ipmitool sdr type current) to env data when available
- added all environmental data to lexpr and sexpr
- added swap total/used to lexpr and sexpr
- building incorrect symlinks to collectl-suse and collectl-debian in INSTALL
- also wrong in collectl.spec
- for IB monitoring, when couldn't find ofed_info was still trying to run it
- need to intialize $interval2SecsReal to i2 first time when 0
- do NOT report process/slab data for the first interval with data in it
3.1.2-1
- more cleanup to INSTALL to give work read access to ARTISTIC, COPYING and GPL
and set a few more protections on other files
- chage to --from/--thru processing since error messages implied you could use
dates too, so now you can. see man page or web documentation on playback for
details
3.1.1-5 November 5, 2008
- two new fields added to slab data to show changes in total allocation between samples
- when mixing --procanalyze with other subsystems, the non-process data wasn't getting
written
- new switch: --slabanalyze
- in header for process data change 'faults are ...' to 'counters are ...' since
we're now including I/O counters as well
- wasn't printing process i/o headers with --procanalyze output. thanks Sven
- when using -on, cpu % needs to divide by the real interval and not 1. thanks to Sven again!
- added percent CPU utilization for process I/O format as well as prc and prcs files
- also --procanalye now honors -om for msec level times
- new process option: c. will include cpu times of any child processes (not threads)
that have since died
3.1.1-4 October 29, 2008
- fixed a rounding problem with numbers between bewteen 1000M and 1024M
that were getting printed as 0G (thanks Marko)
- found error in conversions to K, M, etc where in some cases dividing by
1000 instead of 1024! Specifically: i/o sizes for disk, networks lustre and infiniband.
Also lustre BRW states and some of the KBs fields processes for I/O and memory usage
in the default format - the detailed memory format had it right.
brief formats for: disk, network, quadrics, IB, lustre
But also note these only come into play when values being reported exceed the default
field widths and so ususally aren't tripped.
- changed -oF to --procopts f
- limit username in process display to 8 chars
- make sure terminal echo turned on when falling through error()
- if processes AccumTime>999 minutes in Process Summary, which is pretty rare, drop
fractional seconds resulting in a different format
- included sort-type in top process display
- added ioall to --showtopopts menu
- allow a numeric width to be included with --procopt w
- removed restriction for considering -sl ambiguous and it will be assumed to mean lustre
subsystem rather than a typo for -slab
- misspelled RSys/WSys as RSYS/WSYS in procanalyze code
3.1.1 October 8, 2008
- missing leading space before 'sd' when determining disk names during initialization
can result in wrong devices being listed with -sD and the header if they contain
an embedded 'sd'
- fixed problem with --slabopts s and or S with -P or in playback
- fixed --top checking to verify ALL different I/O related types
- generate an error message for mixing lustre client option O with M or R
- allow printing detaild in --top mode, BUT user needs to control top part of
display with --hr
- playback of environmental data in plot format was printing values every interval rather
than just during interval3
- some impi values may be '' and so report them as 0 to make sure gnuplot can handle it
- make a few changes to INSTALL for debian-based installations
- added 'AccuTime' to top I/O display format
- new feature: top slabs! same switch as top processes, --top, but include names of
slab column to sort by. see --showtopopts
- filtering for old slabs now matches beginning of slab name just like slub
- lustre OST/B data wasn't shifting headers when -oT included
3.1.0 September 3, 2008
- fixed 2 problems in INSTALL (thanks sebastien)
- forgot to copy collect.conf to BINDIR/etc
- forgot to set protections on collectl and inet.d script
- cleaned up interval header printing
- new feature: environmental monitoring via ipmitool
- added environments to daemon defaults in collectl.conf
- changed default interval3 monitoring interval to 2 minutes
- 1st line of brief headers were 1 column too narrow for -t,m,h&f
- when reading lustre MDS stats, don't tell getProc to skip over anything and
save everything that starts with 'mds_'
- extended lustre MDS data reporting
- added I/O size to lustre Client/OST verbose/detail output and make it honor
--iosize in brief mode
- fixed a 1 column formatting shift when using -oT with some lustre client/ost output
- fixed problem in which --hr 1 wasn't causing a new header every intereval for
detail data of same type
- increased size of KBBytes for lustre/interconnect data to 7 digits
- very minor, but if user specified -s+l and --lustsvc and lustre
disabled, only looking at subsys in checkSubsysOpts was generating an error
so now it looks at '$userSubsys' too.
- make $filename local in getSys()
- another pair of switches: -X, --helpall lists ALL help making it possible to
grep for something if you can't remember where it is
- added --grep which allows printing all entries in raw file as timestamped
lines. may mix with other playback switches
- if filtering processes and no data initially collected, interval2Secs will be
0 first time and flt/sec will generate illegal division error so set i2 to 1
- was calling procAnalyze even if no data processed during an interval and as a result
the last pid seen was being credited for that interval when it shouldn't have been
- added parent pid to top i/o display
- when looking for collectl procsses with -sx, be sure to ignore those instances
where the command is 'ssh'
- discovered '$lastInt2Secs' not getting reset when a new set of prefixes
were being played back. This meant the denominator for first line of
process/slab rate data would be wrong, but most people probably wouldn't
have even seen this
- a couple of fixes to correct --procanalyze reporting errors
- removed extra space from --procopts i header.
- significantly expanded --top sort types
- "waiting for..." message will now honor --quiet
- if more than one file played back with interrupt data AND latter one had more
CPUs $intrptLast{}->[$cpu] wasn't getting getting initialized
- allow commas in addition to spaces to separate files in 'playback' list
- discovered a user app can modify contents of /proc/pid/cmdline and so cannot
assume it will always end in null (see test of $cmd1)
- change test of !$slubinfoFlab to $slabinoFlag since both may be missing
3.0.0-4 July 1, 2008
- major switch cleanup
- completed cut-over from -O to xxxopts started in V2.6.4 by creating
--nfsopts/--lustopts. -O kept around for backwards compatibility
for nfs and lustre
- a couple of switch changes to reduce complexity of -o and to clarify new
meaning of handling time offsets and from/thru times for playback
- replaced -ot with --home
- replaced -oP to passwd
- replaced -t/--timezone to --offsettime which now takes a time in seconds
- replaced -b/-e to --from/--thru
- new switch: --procanalyze will produce space separated process summary file
(extension = prcs) that summaries process data for each unique process
- big enhancement for --top. now when -s specified prints a scrolling window
showing histories (-oT recommended but not required) if in brief OR verbose
and all lines the same. note - this mode does NOT support detail subsystem data
- also now identifying the parent who created the thread correctly
- output format cleanup to make things more concise. no changes to plot format
- changed order of columns for brief lustre client to be consistent with all
other brief fields
- changed order of I/O related verbose subsystems (disk, network, infiniband
and lustre) to be more consistent with brief mode. in other words, all input
stats preceed output stats and KBs preceed I/Os. NOTE - the order of the
fields for plot data have not been touched.
- reformatted help to make more readable (I hope) and fit in 80 columns too!
- nfs got inadevetantly dropped as a valid subsystem in V2.6.3 and it's now back
- wrong logic for verifying --procopts Z only allowed in -top mode
- -oA was calling printMini1Counters() instead of printBriefCounters()
- renamed printVerbose() to printTerm() because it makes more sense
- when reading diskstats, make sure leading space before 'sd' as there is
with $diskFilter in formatit.ph
- fixed printing process data that got broken in plot format
- made brief fields 1 column wider for lustre/infiniband in brief mode
- lustre client names didn't make it into header with -sLL was specified using
old option format
- discovered the cvt() routine wasn't being used everywhere in printBrief()
- found/fixed bug that's been there almost forever! if you play back a file
recorded with -sZc but force collectl to only process -sZ it got fatal errors.
Just goes to show how many combinations of conditions there really are!
- fixed problem (I hope) where extra 'RECORD' separators were getting printed
for empty intervals
- fixed code that checks for another instance using IB since it wasn't dealing
with -s using both + and - in it such as a daemon that has -s+YZ-x
- couldn't play back process data on a PC without --passwd since /etc/passwd
not there
- wasn't dividing lustre client OST details by 1024
- discovered/fixed file header entry for switch options which only showed switches
and not options. since a read-only field it shouldn't have hurt anything.
2.6.4 June 11, 2008
- fixed references to gzerror() to be in string context and so error text correct
- miscellaneous documentation changes, mainly to support code changes
- do not report /proc/pid open failures since they happen often enough to be
a nuisance
- changed order of options for --top to be type,num and if no num use the screen
size
- dropped --procio and --procmem replacing them with --procopts i and m
- new options for --procopts: r and z
- broke --vmstat when changed $cls to $clscr
- removed inline code for vmstats since now down via vmstat.ph
- collectl --top generating uninitialized variable message when blank line in
/etc/passwd was fixed
- wasn't honoring -ot for a single subsystem in --verbose mode (sheesh) and now it is
- remove special code that removes collectl from --top display unless explictly
requested. this will help make users more aware of collectl overhead
- found at least one system that returned different format from 'resize' and so
changed pattern match to make it more general
2.6.3 May 12, 2008
- added a README, INSTALL and UNINSTALL to the tarball to aid in manual
installation and removal
- changed --procopts to --procfilt and --slabopt to --slabfilt because I want
to differentiate between options and filters.
- enhanced socket error handling
- new I/O output data for disks, networks and interconnect
- i/o sizes will always be included in verbose output
- new switch --iosize will add to brief displays
- NOTE this data is not written to tab file since it can be derived
- changes to -si (inode data)
- removed info from header and will get it from proc instead
- changed what is reported as some fields no longer valid and added 'number' of
dentry noting that the values for 'unused', which increase as files are
created makes no sense to me. also including file handles and inode counts
in brief format.
- as a result of adding -si to brief format, --all results in brief output for
everything and so you'll need to include ---verbose to see verbose form
- several new options for --procopts (thanks for the push Matt)
s: will add read/write system calls to process stats
t: will force collectl to look/display threads for ALL processes
note that this can be a lot of overhead if there are a lot of threads on
your system. All you threads can also be seen via 'ps -eLf'
w: will make display wider by including arguments to process names
- you can now request what to sort on for --top (cpu, io or page faults)
- you can now include --procfilt with --top and it will only consider
those processes that match for display
- you can now use --top in playback mode
2.6.2 Apr 29, 2008
- forgot to rename call to resetMini1Counters() in collectl.pl
- do NOT clear $miniDateTime when --export
- added swapin/sec and swapout/sec to [MEM] data in tab file
2.6.1 Apr 24, 2008
- for perl version checks, use 2 digit minor/patch levels (thanks devzero)
- report zlib and HiRes vesions in collectl -v output
- grab ALL of /proc/meminfo for non-2.4 kernels even though we're not
processing all of it
- added the number of active lustre file systems seen by the client for lexpr/sexpr
- was incorrectly restricting -A to -P or --export and that was wrong
- allow --export in playback mode, making it possible to use --vmstat as wellx
- extended --top to allow -s to be included along with proc stats. not that pretty
but very useful
- renamed printTerm() to printVerbose(), briefFormat() to printBrief() and other
associated printMin1 routines
- when changed syswrite() in writeData in last version, lost trailing /n and
so put it back
- ibcheck was redefining global $port so reopening socket in 'server' mode failed!
- when --export added forgot to handle writeData() conditional correctly for
process and slab data
2.6.0 Apr 03, 2008
- lustre
- typo for lustre readahead 'not consecutive' variable!
- added 2 new readahead variables for 'failed grab...' and 'wrong page...'
- extended meaning of --headerrepeat and added a synonym of --hr for it
a value of -1 means never display a header and 0 means only display it once,
eliminating the need for -oH and -oh which are still supported but not shown
in help. They will be eliminated in a future release.
- bug in regx prevented gzclose on zipped tab file
- cleaned up code (finally) that deals with displaying headers such as how often
and when to skip entirely. this included dropping the -oh option which
predates --verbose mode
- if we can find 'resize', use it to get number of lines in display and use for
default. This can still be overriden in collectl.conf
- slight change to -ot behavior. only erase screen one time and then just
overwrite what's there as it's softer on the eyes
- fixed format error in 's-expr rate' for disk summary stats
- modification to the way --custom is used to make it work with -f, -P, sockets
and --rawtoo just like --sexpr. In fact, sexpr and vmstat code has been
removed from formatit.ph and are now standalone include file named sexpr.ph
and vmstat.ph respectively. See documentation for more details.
- renamed --custom and --custdir to --export and --expdir to better reflect
that the main purpose of these is to export data to a file or over a socket
- had to move subsys/interval initialization code around to happen before
calling --export
- changed init.d file for SuSE as it couldn't detect collectl running
when pids was 5 digits
- added code to handle write of partial data over socket
- based on popular demand, --all has been provided to show all summary stats.
be sure to try it with -ot
- added CPU number to process detail report. since this data has actually been
collected all along, you can play back older raw file and now get them
- changed default socket port to 2655
2.5.1 Mar 21, 2008
- added OFED 1.3 location for perfquery to collectl.conf
- added new constant for ofed_info to collectl.conf
- if can't find perfquery and/or ofed_info, ask rpm and if there update collectl.conf
- redefined debug flag of 8 for lustre checks and leave 2 for interconnect only
- adding more debugging details for infiniband initialization
- changed daemon startup switches to include -sC. this will NOT generate any extra
load on collectl but will cause CPU details to be generated in plot format which
will include interrupts/cpu
- make sure user have privileges to run perfquery
- moved location of --sexpr with -sj check
- for lustre versions < 1.6 don't limit BRW stats to being in directory with
MNT in its name, which was certainly the case for HP-SFS
- changed headers for lustre rpc buffers to 'P' rather than 'K'
- changed directory on MDS that we look in for stats from .../MDT/mds/stats for
older versions of lustre to ...MDS/mds/stats for versions >= 1.6
- need to check lustre version BEFORE calling lustreCheck() routines
- in lustreCheckClt(), only do OST level tests if really a client
2.5.0 Feb 29, 2008
- if HCA present but IB stack not completely loaded, the cat of /sys/class/infiniband/*
fails and reports error. redirecting STDERR supresses that error
- added support for reporting interrupts by CPU
- removed all but the collectl and collectl-data man pages, moving their content
to the collectl web site at sourceforge AND to /opt/hp/collectl/docs
- when installing is a brand new ROCKS environment /bin/rm not there yet so make
conditional in %pre section of spec file [thanks roy]
- modified spec file to add build level to release so I can keep the release number
the same [thanks again, roy]
2.4.3 Feb 04, 2008
- cpu percentages calculations need to include iowait in denominator
- memory stats: include AnonPages in mapped memory
- fixed pattern match for IB device number to properly select mlx4_ adapter
- was incorrectly including network bond stats with total network stats
- wasn't printing date/time for --vmstat when requested
- added IbDupCheckFlag to collectl.conf to allow disabling the check for
duplicate instances both trying to read IB counters
- removed a couple of spaces from default output so now <80 columns wide
- when someone creates a new logical disk after collectl has been started, we need
to add that disk to the list of valid disk names
- changed the algorithm used to check for bogus network data. you can also
disable these checks by setting DefNetMax to a negative value in
collectl.conf
2.4.2 Jan 16, 2008
- changed purge algorithm to explicitly purge any files in the logging directory
that match hostname, contain date/time stamp and do NOT end in 'log'. Before only
raw files were purged and this was clearly not the intent.
- on a lustre MDS, the mds_sync counter has moved as well as others added so pull more
of them. even though the newer ones won't be reported on, they'll be in the 'raw'
file for reference via tools like grep.
- bogus network record processing changed as follows:
- use double the reported network speed from the raw file header and if not known use
the DefNetSpeed in collectl.conf which for now it 10000Mb.
- was setting 10G network speeds in header wrong (wasn't multiplying by 1000). Therefore
if we find a network with speed of '10' on older version multiply by 1000 as this only
effects 'bogus' check.
- added IB speeds to network interfaces in header BUT limited to OFED and assuming
all devices running at same speed
- looks like I broke old style slab reporting! the data is still collected correctly
but won't print. now it will...
- just discovered you couldn't print to terminal in plot format for -sY or -sZ, though I
don't know why you would ever want to! in any event, they now call writeData() and so can...
- if writing to non-compressed files and the flush time is less than the interval, just
open the files with autoflushing enabled to save flushing overhead
- plot format for [SOCK] data was sticking an extra $SEP in header after [SOCK]Tw
2.4.1 Jan 05, 2008
- corrected calculation for cpu times to include soft, irq and steal which was
causing incorrect values to be reported for system with higher values in one
more of these counters. If one replays any existing raw files the correct
values will be produced.
- added support for new SLUB slab allocator which results in different output
format for slab reporting.
- added 'Flags:' to header and use value if 's' to indicate a raw file contains
new slab data and a 'i' to indicate that process data contains I/O counters
- also added slab alias names as a block comment directly below main header
because they are needed for playback on a different system or even on the
same one in case the slab configuration has changed
2.4.0 Dec 23, 2007
- test for -f filename only checked for existing directory and if ended in /
still created it but logfile created started with -
- changed way lustre versions and services formatted in header because when used
very old file where neither cfs or sfs version defined we get an extra CR which
screws up gnuplot. I'm probably the only one who will even see this.
- typo prevented OST BRW stats headers from printing properly which in turn messed
up plotting
- remove arg-list from process command string when displaying in terminal format
- added process i/o stats for systems with that feature built into the kernel
- include new flag --procio which functions silimarly to --procmem in that it
show much more detail about stats
2.3.4 Dec 13, 2007
- added IB code of 0c06 for Mellanox IB Infinihost III card
- expanded --sexpr behavior to allow sending over a socket or even to
stdout and for consistency, logging to a local file is no longer
required, though logging is certainly permitted and as a result
more consistent. the collectl-logging man page has been modified
to address this
- forgot to add cpu irq, soft and steal to sexpr header and raw routines
- removed -H as it's no longer needed given all the other data export options
but preserved it's functionality by redefining the meaning of -d4 and -d32
2.3.3 Oct 16, 2007
- added 3 new fields to CPU values -- irq, soft and steal, which resulted in
a change of order of the verbose output, the theory being it's more
important to have the display in a readable order rather than just append
the fields to the end. While at it the same was done to ALL most CPU
output formats for consistency
- incorrectly included !$plotFlag in test to set $zFlag and as a result
'flush' wasn't working for raw files
- extended IB interface support to include ConnectX mlx4
- when printing 'brief' subtotals for infiniband, do not average errors since
intermittent error rates may be too small to see so just print increasing
totals
- rare case - if doing slab/proc and we only specify int 2 AND less than
default interval (say we do -i:.1), we need to force int1 to be int2 so
we don't get error that int2),
changed context switches and interrupts to show averages rather than totals
- subtotals in brief format were including incorrect units (K,M,G) for some
fields including disk, network and lustre
- converted string to seach for 'vstat' from a single bin name to mulitple ones
and updated processing accordingly
- if specifying -l with o/m on lustre client was incorrectly checking for disk stats
file which it should only do if -OD specified
1.7.4 Jun 05, 2006
- whether someone requests -sc or -sC, collect both types of stats. this
makes it possible to play back either independent of what was specified
at collection time making CPU stats consistent which others which also
exhibit this property
- -V now shows interactive and daemon defaults separately
- include the name of the host running collectl in the logfile name
for uniqueness
- if corrupted file, include name in error message AND if multiple files being
processed, skip remainder of currupted one and go on to the next one
- changed sfs readhead in brief mode from hit precentage to show actualy hits/misses
as these are so typically close to 05 or 100% you can miss the changes
1.7.3 May 23, 2006
- the plot format headers I thought got released in 1.7.2 didn't. they do
in this release
- if user had specified lustre options and lustre wasn't active, the playback
fails because the header says there isn't any lustre data present but -s says
this is. the fix is to recognize this condition during playback and simply
ignore the options
1.7.2 May 17, 2006
- added qualifiers to plot-format headers to clarify which subsystem
the data relates to
- changed headers for lustre -OB output from K to P since the data being
reported IS in pages, not KB
- commented out the check that limited the data being reported for disk stats
from 512KB to the full range of up to 2MB
1.7.1 May 03, 2006
- make sure -M1 and -oh turned off in -H mode
- support for new /proc format with elan 5.20
- fixed a bug that prevented elan detail files from being written to
- wasn't building correct link in /etc/init.d for suse/debian
- changed rules for determining hyperthreading: true if siblings/cores==2
- added cpu vendor, speed, cores, sibings to common header
- added support for 'official' stats from Voltaire so if
/proc/voltaire/adaptor-mlx/stats exists, it looks in there. otherwise
it tries to use /proc/voltaire/ib0/stats. note - since voltaire only
supports a single HCA, if more than one is found only the first it
looked at and an appropriate warning generated
1.7.0 Apr 21, 2006
- fixed bug in thread process reporting. although the treads numbers were
getting reported correctly, all lines were reporting the parents stats
- changes to -O to include lustre options broke some nfs options
- also updated process man page to more clearly articulate what gets reported
- need to make sure raw entries that start with 'Slab' are not coming from
/proc/memory by making sure no ending ':'
- there were problems running on SFS Admin node when -sl specified because of
logic error dealing with monitoring a service not yet up
1.6.9 Apr 02, 2006
- when specifying -i:xxx in interactive mode, default monitoring
inteval was not changed to 1. this effected displays for slabs
- verbose mode wasn't working correctly
1.6.8 Mar 28, 2006
- fixed a few minor problems to ensure works with colplot/colgui
- reformat FAQ and include in kit
1.6.7 Mar 20, 2006
- didn't quite get -s right for overriding that in file
- made size of disk block buckets bigger but limiting the size if
displayed values based on collectl.conf entry
- OST detail detail data incorrectly written to CLT detail file (which
usually isn't even opened!)
- remove default of -oT, it complicates things...
1.6.6 Mar 07, 2006
- added Subsys to 'RECORDED' section of header (note spelling to preserve
pattern patches for second occurance) and also updated Subsys data to
correctly reflect values based on +/- in -s during playback
- removed erroneous check that prevented running -sL -OB on lustre OSSs
- forgot to write headers for OST detail data
- remove test for ignoring -sL for mds data since we now CAN have some,
but remember that it goes into the .blk file
1.6.5 Feb 28, 2006
- verify system is an mds/ost before trying to open block iostats file
- when lustre subsystem disabled wasn't removing 'l' from $subsys
- incorrect check for valid -M1 subsystem in setOutputFormat(). was
invalidating any uppercase letters and show have only looked at
$MiniSubsys
- changed why $BinDir gets built to work with more complicated links
- missing cvt() in print for lustre client summary
- was incorrecty setting reportOstFlag to zero when -oD and supressing output
on playback
- changed -o^h to -o-h
- removed support for -t as it's not really useful and anyone using it
(and I doubt there any) don't really understand -P
1.6.4 Feb 25, 2006
- default settings for non-daemon mode are not '-i1 -scdn -M1 -oT'. furthermore
if single subsystem -oh is also set. you to remove -oT or -oh, preface that
switch with a ^ such as -o^T.
- added -S switch, which means collectl was started remotely by something
like 'ssh' or 'rsh'. at the end of each collection interval see if
parent daemon when away and shut down...
- when missing HiRes time module and -om or fractional intervals specified,
ignore -om and round off interval
- wrong error message about requiring HiRes for -om as it reports
-P is required instead
- removed -OB processing code from ministat processing
- produced uninitialized errors if nothing selected during playback do to
incorrect settings of -b or -e. now generates an error.
- only write inode info to header if -i
1.6.2 Feb 01, 2006
- fixed bug in ELAN monitoring logic that was causing it to create new
logfile every 15 minutes
- fixed bug which generated unitialized variable in some cases when
slab and non-slab data with -M1
- replaced 'cat' command in routines that check lustre/interconnect state
to cat() which is more efficient
- fixed bug that prevented raw file with no core data from being played back
- added support for lustre client rpc and readahead stats.
see man collectl-lustre for details
1.6.1 Jan 23, 2006
- changed using units for ELAN stats to KB to be consistent with IB stats
- added new option to -M1 mode such that is a user types A, the averages
will be displayed
- added new switch (groan) -oA, which when playing back a file in -M1 mode
will append Averages/Totals
- only use ethtool to determine network speed if root
- playback error reporting was not explict enough when wildcarded
specification didn't match anything
- moved around code that expands -s (based on +/-) so expanded value
seen with -d4096
- infiniband support - requires special internal '/proc' module
- summary data changed for quadrics (including plot format) so that it
matches that of IB and any potential future interconnects to only show
total errors and not individuals counts which IS available as details.
This was necessary so that plotting tools can display interconnect data
independent of its type.
- incorrectrly determining kernel version if version containg 2.4 in interior
of id string
1.5.8 Dec 14, 2005
- add additional copyrights to source and manpages
1.5.7 Dec 08, 2005
- added more conditional execution if not pcs for things like `date` and
existence of lspci and ethtool
- add setsid to deamon startup and reset terminal I/O channels to /dev/null
in child
- entire chunk of header line and data that followed not getting prepended
with hostname when -A and -M1
- only complain about missing lspci or ethtool when not in playback mode
- need to determine path to collectl using readlink() in case defined as
a link
- latest version expanded size of buffers to 8 so had to modify display
- added a warning if new network device found after started, which will cause
uninitialzed variable warning. this may go away when started at S99
- renames release notes so they can co-exist with release notes from other
tools
- modifications to 'spec' file
- deamon will now start at S99 to give more devices a chance to initialize
and be seen
- install in /opt/hp/collect, link to it from /usr/sbin
1.5.6 Sep 23, 2005
- modified data collection for 2.6 memory to include everything up to Vmalloc
- extended -sm reporting to include slab and mapped memory for all outputs
- was not printing header for first time for -M3
- check for non-existant /proc for inode processing
- remove debug check for open proc error messages
- fixed bug in collectl.conf. setting Interval2 overwrote Interval.
- debug flag of 4096 prints header
- make sure under -i0 sampling that the number of intervals for processes and
environmentals are proportional to their default timings
- removed 'cciss/' from disk names in file headers
- moved socket handling code to the front so anyone who calls us and gets
an error or does a 'collectl -v' will see the socket open and then close
so it can then cleanly exit.
- added Swap size to header
- if ethtool present, record 'eth' speeds in header
- generate an error if lspci not on system.
- warn that no eth speeds in header it no ethtool on system
- inserted inter-file marker into PRC files so can differentiate between
processes with same pid/name from different logs
- printing wrong header for lustre client details in plot formatit
- allow printing process data with date/time stamps so when you grep the
output you can see them
- fixed erroneous message "Looks like 0 exited so not looking for new
threads" which should not be reported when value is 0.
1.5.5 Sep 01, 2005
- added exception processing for lustre client summary data
1.5.4 Aug 31, 2005
- added exception reporting for lustre KB/sec read/write for OSS and
Reints/sec on an MDS. For now it's NOT writing to an exception file
as I'm thinking of removing that capability since I don't believe it
is used very often AND there are too many files written already!
1.5.3 Aug 29, 2005
- serious bug fixed. when playing back any files and producing process or
slab files (.prc or .slb), ALL other data was being skipping for that
period. The workaround is if you need both process/slab and other data
you'll need to do it in 2 batches!
- very minor (but annoying). if interval ends in exactly .000 seconds,
$seconds is treated as an integer and splitting on '.' provided an
undefined $usecs which in turn generates an uninit var at line 2456
- skip over IB support since net yet fully baked
- the line RECORD... was leaking through without a hostname prefix with -A
- flush mechanism which was incorrectly flushing every interval
- internal coding thing: cleaned up reporting to be driven off $subsys
and collection driven off the 'flags'. this means that doing -sl -L in
record mode will not display lustre stats on a non-lustre system as is
already done in playback mode
- check for location of lustre modules expanded to support newer releases
- set ALL output files to autoflush on write when printing in plot format
on the terminal. not doing so was causing collectl to lock up until the
output buffer filled when called from a script with -oh or -oH
- InfiniBand Support
- changed method for determining lustre driver installed. now just looks for
anything named 'lustre' in /lib/modules
- updated man page to note that some subsystems, specifically d, l, n, t, x, y
thought recorded in summary mode CAN be played back in detail mode and visa-versa
- added a restriction that you can't play back a file records with -sd using -sD
if it wasn't originally recorded using -sc as well (need times in jiffies for
some calculations)
- added additional parameter to collectl.conf to point to additional library paths
(primarily for development but may prove useful later on)
- made some changes to header
- added seconds associated with timestamp of filename timezone
- renamed 'Daemon Options' to 'DaemonOpts'
- moved some fields closer together
- added a preamble for plot format files that show original collectl version and
switches
- added HiRes flag state of original collection so playback knows
- Another switch -T to control time zone conversions on playback, which is
required for dealing with files that don't have enough info in header to
to autmatic conversion of times
1.5.2 May 23, 2005
- removed extra comma in printf statement at line 3752
- warning about -sL and MDS was missing 'if ...' modifier
- need to flush buffers before creating new logs
- not declaring $datetime as 'local' was generating unit vars with -M1
- the pattern match on sd disks wasn't set to pick up disks with 2 alpha
chars after the initial 'sd'. it does now.
- print error messages to terminal via STDERR
- moved location of slab/proc initialization to record mode as it was causing
problems on PCs.
- when no process data exists, print 0 instead of '-'
1.5.1 May 03, 2005
- updated several man pages and created a new one: collectl-lustre
- added memory size to file headers
- strip any quotes from playback file name that may have leaked in
- if the destination directory doesn't exist, create it
- some Linux specific code was moved/modified to facilitate running on a pc.
- initialization of $MyHost and $OS.
- only call syslog on linux and so we have to do a 'require', not 'use'
- a number of changes to support dynamic identification of lustre
configuration changes
- change lustre related information in file headers
- no longer an error to request -sl when no services present
- printing lustre client data in plot format was missing 2 fields
- replaced common '.lus' detail file with specific ones of '.ost' and '.clt'
- KNOWN PROBLEM identified with lustre client data collected using -sLL with
older versions
- KNOWN PROBLEM identified with lustre client size read/write I/O counts
- modified recognition of quadrics such that if /proc structures are present but
driver not loaded, a warning rather than an error followed by an abort occurs.
- new switch: -V to print operational defaults
- storing kernel version rather than whole o/s name in header
- only include SCSI info in header if non-blank
- do not print header when plot output directed to terminal
- added '[HYPER]' to cpu display header when cpu hyper-threading on
- expanded scope of -m to print playback processing messages on terminal
- problem writing to syslog on some systems and so that function currently
disabled
- made width of network name dynamic to account for IB names
1.3.2 Mar 13, 2006
- for M3 reporting, headers weren't printing
- /proc/slabinfo went to V2.1 with no format change, so had to extend
the check to include anything in 2.*
1.3.1 Jan 18, 2005
- problem corrected with slabs/pagesize during playback
- when a new slab was created after collectl started, during playback
an 'uninialized variable' message was being generated.
- fixed problem with -b/-e when date specified
1.3.0 Jan 18, 2005
- write startup/shutdown messges to /var/log/messages when writing
to files (too much of a nuisance to do for all invocations).
Write ALL fatal errors to messages. This is in addition to the
normal logging that gets written to collectl's own message log
in the logging directory, which is only written to when writing to a file.
- installation no longer saves old startup script since user customizations
haven't been there since introduction of /etc/collectl.conf
- support for SuSE distro based installs
- support for Debian installs as long as one converts rpm to deb
- new man page for process monitoring and how the math works
- moved data definitions and examples to their own man pages
- make process sort order ascending numeric
- added '+' option to -Z switches which results in threads being displayed
- see man page for restrictions
- -Z enhanced to allow filename to be specified as an alternative. see -h
- added startup switches to log header (don't know why I didn't think of earlier!)
- expected for lowercase hostname in playback file. this was a bug!
- removed partition specific code and now share 2.6 /proc/diskstats
- removed -spP and collectl uses nows /proc/partitions if it contains data
- removed -oP as it was never fully debugged and buggy
- slight format change of disk stats output to make consistent across 2.4/2.6
- fixed bug when specifying a playback file wildcarded spec that didn't
start with full hostname
- fixed bug when playing back multiple files from multiple dates with
-b/-e switches
- found bug in the way linux handles reading /proc - may read past end of
existing structure) and so changed handling of /proc/pid/stat to only read 1 line
- make sure collectl will run on windows in playback mode by changing some
linux specific code
1.2.5-4 Dec 14, 2004
- modified processing of /proc/net/netstat to accomdate slightly different
format with debian 2.6 kernel (leading blank line not expected!)
- removed 'partition' as valid subsystem for 2.6 kernels as that data now
comes from /proc/diskstats
- wasn't recognizing 'sd' devices in /proc/diskstats
1.2.5-3 Dec 14, 2004
- found latent bug - thanks tom - in process processing code. when not using -Z
new processes weren't discovered. now they are!
- had to move 'interval' processing code to section before alarm set
- wan't honoring -t when printing process data
- on systems where lustre fs was created outside of 'sfs' environment. MDS
directories had different name format so pattern match had to change
- not all MDS read/write fields always defined and so conditional prints req'd
- changed single quotes in man page to \` so at least something would print
- removed 'C' as a valid option which was moved to -O a number of versions ago
1.2.5-2 Dec 14, 2004
- fixed a problem in 1.2.5 which wasn't properly handling return status from gzflush()
- enhanced error messages when invalid subsystems are specified with '-'
- removed restriction against adding core subsystems with '+'
- enhanced -sLL processing to deal with /proc with data in different positions
- bumped indexes for getProc() 12/13 to 13/14 to keep Lustre processing together
- if flush timer set AND using -H, only execute command every flush interval
- flush timer was off by 1 second (tested it using > instead of >=)
1.2.5 Nov 03, 2005
- The 'subtotal' feature of -M1 causes cron based scripts to blow up because
the 'M1' code wants to check for terminal I/O. This feature has now been
disabled for environments with no terminal.
- Found a corrupted compressed raw file! Closer inspection of collectl
showed no error handling for gzflush() errors and limited error handling
for gzwrite errors. Changed to close/recreate new logs on zlib errors or
abort if recovery impossible. As a safety net will kill itself if there are
ever more than $ZlibMaxErrors in a single day - currently setting that value
to 20 but can be overridden in collectl.conf.
- Added code to make sure valid time marker in raw file and if not to declare
file corrupted and exit. This is for the case noted above where a compresses
file had bad data in it. Still a mystert how this happened but I'm hoping
it was related to gzflush and so now shouldn't happen again.
1.2.4-3 Oct 19, 2004
- _SC_PAGESIZE posix variable not supported on 5.6 releases of perl and
results in uninitialized variable warning and errors during SLAB
reporting. Added code to force pagesize to 4096 for IA32 and 16384
for other architectures which is not completely correct for all cases.
1.2.4-2 Oct 13, 2004
- using wrong pagesize for slab calculations on IA64. use sysconf()
and added PageSize to logfile headers
- writes slab version into header instead of whole version line and drive
format off that number. report errors for unsupported versions.
- removed a line of code that couldn't execute if daemon found to be
already running
- found reference to cvt2() which was removed in last version
- ministats with a -c leaving echo turned off!
- changed width of memory reporting fields for slabs and memory stats
from 6 to 7 so more significant digits retained when displaying a
number like 123456K which in 6 columns displayed as 123M. Not sure
if it should be done to other fields as well because it does effect
screen real estate. Remember to use -w to get rid of K/M/G
- removed unused c and s from -O
- added code to filter slabs and remove those with no allocations,
-Os, or those with no change in slab activity since last interval, -OS.
NOTE - slab objects change all the time and so including them in the
filter is pointless
- added 't' to core variables to be monitored since it only takes about
an extra cpu second per 8640 samples.
- added 'sockets' to ministats. this means one can now simply do
'collectl -M1' and get ALL ministats for default variables (of course you'll
need a VERY WIDE window)
1.2.4 Oct 07, 2004
- added SLABS!
- add -oF tell collectl to use cumulativetotals for Maj/Min faults
- fix printing of process data in plot format
- found a couple of inconsistencies when reporting in 'K/M/G' format.
made sure bytes /1024 and counts/1000
- fixed uninitialize variable with -sP -p xxx
- echo not turned back on with -M1 and -p
- fixed bug with -p -f -P
- changed $ProcInterval/$EnvInterval names to $Interval2/3 so that
slab interval processing could be slipped into $interval2. this is
really an internal thing.
1.2.3 Sep 18, 2004
- Added -sZ and -Z to capture process data.
- Added -M3 to report finer detailed memory data for processes
- Added ability for dynamic subtotals with -M1 (see manpage)
- Added -st and -sT for tcp counters
- Made lustre and quadric counters part of default subsystems
1.2.1 Aug 12, 2004
- Support added for lustre clients
- Search for 'collectl.conf' in /etc, collectl bin dir, then current
dir if no -C
1.2.0 Jul 16, 2004
- Support added for quadrics and lustre
- Replaced /usr/sbin/collectl.ph with more flexible /etc/collectl.conf
- Added switch to override location of /etc/collectl.conf
- Performance improvements to /proc processing
1.1.13 Jul 16, 2004
- Fixed timer bug (since there since version 1) that only shows up
on 2.6 kernels
- Fixed formatting bug in NFS detail display introduced by -oT
- Preserve /etc/init.d/collectl so any custom changes are preserved
across versions
1.1.12 Jun 07, 204
- Fixed bug that was causing nfs client data to not be correctly captured
- Moved nfs client option 'C' from -o to -OP/dd>
- Changed format/use of -M1: more/optional fields
- Made timestamp line formats dDT available for 'standard' output
- Changed column widths for DISK/PARTITIONS fields from 4 to 6 as needed
1.1.11 May 21, 2004
- assured 'wc' used correctly since 2.6 kernels changed format
- renamed kit to 'noarch'
1.1.10 Jul 16, 2004
- Fixed extra space getting printed in plot format by -ss.
- removed -a 0 from init.d file since it would prevent starting on
machines that don't have HiRes installed
1.1.9 May 03, 2004
- The biggie - support for 2.6 kernels resulting in changes to -sd & -sp
- Added -a to startup script to align times to minute boundary
- New switches for ministats to control date/time format: -o dDT
- Combined ministats 1 thru 4 to more intelligent -M1
- New ministat M2 mimics vmstat but with date/time stamps
- Clarification of memory statistics in man page
- Minor bug fixes with custom ministat directory name processing
1.1.8e Mar 24, 2004
- Added new subsystems -sE -slL [environmental and lustre]
- Added new output options -otH
- Added 'ministats' which are combined subsystems on singe line
(see manpage for -M)
- Added ability to send output to a socket for remote monitoring
1.1.7 Feb 06, 2004
- Updated Copyright notice
- Fixed bug when trying to use -P without -f
1.1.6 Dec 05, 2003
- wan't properly handling -p -f -P for multiple files on same day but
different -s values: first one overrides the rest!
1.1.5 Dec 03, 2003
- added error checking to ignore partially read /proc data
- added ability to specify YESTERDAY or TODAY in playback file name
1.1.4 Nov 26, 2003
- bug in playback of nfs client data
- enhance -v to show zlib/compress if present
- fixed bug handling -p -f of mulitple files on same date
- fixed but handling logs from different systems with different subsys values
1.1.3 Nov 26, 2003
- added support for Smart Array devices in partition table reporting
1.1.2 Nov 26, 2003
- added support for disk and partition exception reporting along with
several switches to support it. See -l, -L and -o x/X
collectl-3.6.9/hello.ph 0000644 0001750 0001750 00000014324 12230241726 013135 0 ustar mjs mjs # copyright, 2003-2009 Hewlett-Packard Development Company, LP
# H e l l o W o r l d
# Though not required, it is especially useful to use strict to force all variable to be
# declare and minimize the risk of stepping on any that collectl itself uses
use strict;
# Allow reference to collectl variables, but be CAREFUL as these should be treated as readonly
our ($miniFiller, $rate, $SEP, $datetime, $intSecs, $showColFlag);
# Global to this module
my $counter=0;
my ($hwOpts, $hwTot, @hwNow, $hwTotTOT);
# support for 's','d', and 'sd, assuming 's' is the default. Since collectl doesn't
# restrict which options are valid, we must to it ourself and call errror() accordingly
# For an additional example of error handling, if this module required CPU data to be
# collected as well, you could incldue an error message based on the condition $subsys!~/c/i;
sub helloInit
{
my $impOptsref=shift;
my $impKeyref= shift;
$hwOpts=$$impOptsref;
error('valid hw options are: s,d and sd') if defined($hwOpts) && $hwOpts!~/^[sd]*$/;
$hwOpts='s' if !defined($hwOpts);
$$impOptsref=$hwOpts;
$$impKeyref='hw';
return(1);
}
# Anything you might want to add to collectl's header.
# Try the command 'collectl --import hello --showheader'
sub helloUpdateHeader
{
my $lineref=shift;
$$lineref.="# HelloWorld: Version 1.0\n";
}
# Simulate 3 lines of data being read from /proc and include a further qualifier to act
# as a device number. See how this is used in Analyze().
sub helloGetData
{
for (my $i=0; $i<3; $i++)
{
my $string=sprintf("HelloWorld %d\n", $i*10*$counter++);
record(2, "hw-$i $string");
}
}
# Reset running total for the 3 'devices' for the current interval, which will be
# diplayed in both brief and summary formats.
sub helloInitInterval
{
$hwTot=0;
}
# We could get fancier and look at how much each counter changed between intervals by
# subtracting the last value from the current one, but we're only going to look at
# explict values to keep things simple.
sub helloAnalyze
{
my $type= shift;
my $dataref=shift;
$type=~/^hw-(.*)/;
my $index=$1;
my @fields=split(/\s+/, $$dataref);
$hwNow[$index]=$fields[1];
$hwTot+=$fields[1];
}
# This and the 'print' routines should be self explanitory as they pretty much simply
# return a string in the appropriate format for collectl to dispose of.
sub helloPrintBrief
{
my $type=shift;
my $lineref=shift;
if ($type==1) # header line 1
{
$$lineref.="<-Hello->";
}
elsif ($type==2) # header line 2
{
$$lineref.=" Total ";
}
elsif ($type==3) # data
{
$$lineref.=sprintf(" %4s ", cvt($hwTot/$intSecs));
}
elsif ($type==4) # reset 'total' counters
{
$hwTotTOT=0;
}
elsif ($type==5) # increment 'total' counters
{
$hwTotTOT+=$hwTot;
}
elsif ($type==6) # print 'total' counters
{
# Since this never goes over a socket we can just do a simple print.
printf " %4s ", cvt($hwTotTOT);
}
}
# The only magic here is knowing when to print a headers. Note the use of $rate which
# you can see change with you use -on. Since all -on does is set $intSecs to 1, there's
# no custom coding required. Also note how $datetime and $miniFiller are used together to
# allow the actual timestamps to align with the header correctly.
sub helloPrintVerbose
{
my $printHeader=shift;
my $homeFlag= shift;
my $lineref= shift;
# Note that last line of verbose data (if any) still sitting in $$lineref
my $line=$$lineref='';
if ($hwOpts=~/s/)
{
if ($printHeader)
{
$line.="\n" if !$homeFlag;
$line.="# HELLO STATISTICS ($rate)\n";
$line.="#$miniFiller Total\n";
}
$$lineref.=$line;
return if $showColFlag;
$$lineref.=sprintf("$datetime %7s\n", cvt($hwTot/$intSecs,7));
}
$line='';
if ($hwOpts=~/d/)
{
if ($printHeader)
{
$line.="\n" if !$homeFlag;
$line.="# HELLO DETAIL ($rate)\n";
$line.="#$miniFiller HW Value\n";
}
$$lineref.=$line;
return if $showColFlag;
$line='';
for (my $i=0; $i<3; $i++)
{
$line.=sprintf("$datetime %2d %7s\n", $i, cvt($hwNow[$i],7));
}
}
$$lineref.=$line;
}
# Just be sure to use $SEP in the right places. A simple trick to make sure you've done it
# correctly is to generste a small plot file and load it into a speadsheet, making sure each
# column of data has a header and that they aling 1:1.
sub helloPrintPlot
{
my $type= shift;
my $ref1= shift;
# H e a d e r s
# Summary
if ($type==1 && $hwOpts=~/s/)
{
$$ref1.="[HW]Tot${SEP}";
}
# Detail - these typically have :devname inside the []s
if ($type==2 && $hwOpts=~/d/)
{
for (my $i=0; $i<3; $i++)
{
$$ref1.="[HW:$i]Val$SEP";
}
}
# D a t a
# Summary
if ($type==3 && $hwOpts=~/s/)
{
$$ref1.=sprintf("$SEP%d",
int($hwTot/$intSecs));
}
# Detail
if ($type==4 && $hwOpts=~/d/)
{
for (my $i=0; $i<3; $i++)
{
$$ref1.=sprintf("$SEP%d$SEP%d", $i, int($hwNow[$i]/$intSecs));
}
}
}
sub helloPrintExport
{
my $type=shift;
my $ref1=shift;
my $ref2=shift;
my $ref3=shift;
my $ref4=shift;
my $ref5=shift;
my $ref6=shift;
if ($hwOpts=~/s/)
{
if ($type eq 'l')
{
push @$ref1, "hwtotals.val";
push @$ref2, int($hwTot/$intSecs);
push @$ref5, 1; # makes it a gauge and so an avg for 'tot'
}
elsif ($type eq 's')
{
$$ref1.=sprintf(" (hwtotals (hw %d))\n", int($hwTot/$intSecs));
}
elsif ($type eq 'g')
{
push @$ref1, "hwtotals.hw";
push @$ref2, 'num/sec';
push @$ref3, int($hwTot/$intSecs);
}
}
if ($hwOpts=~/d/)
{
if ($type=~/[gl]/)
{
for (my $i=0; $i<3; $i++)
{
if ($type eq 'l')
{
push @$ref3, "hwinfo.hw$i.val";
push @$ref4, int($hwNow[$i]/$intSecs);
}
else
{
push @$ref1, "hwinfo.hw$i.val";
push @$ref2, 'num/sec';
push @$ref3, int($hwNow[$i]/$intSecs);
}
}
}
elsif ($type eq 's')
{
$$ref2.=" (hwinfo\n";
$$ref2.=" (name 0 1 2))\n";
$$ref2.=" (hw0 $hwNow[0])\n";
$$ref2.=" (hw0 $hwNow[1])\n";
$$ref2.=" (hw0 $hwNow[2]))\n";
}
}
}
1;
collectl-3.6.9/formatit.ph 0000644 0001750 0001750 00001310030 12230241726 013651 0 ustar mjs mjs # copyright, 2003-20012 Hewlett-Packard Development Company, LP
#
# collectl may be copied only under the terms of either the Artistic License
# or the GNU General Public License, which may be found in the source kit
# local flags not needed/used by mainline. probably others in this category
my $printTermFirst=0;
# shared with main collectl
our %netSpeeds;
# these are only init'd when in 'record' mode, one of the reasons being that
# many of these variables may be different on the system on which the data
# is being played back on
sub initRecord
{
print "initRecord() - Subsys: $subsys\n" if $debug & 1;
initDay();
$rawPFlag=0; # always 0 when no files involved
# In some case, we need to know if we're root.
$rootFlag=`whoami`;
$rootFlag=($rootFlag=~/root/) ? 1 : 0;
# be sure to remove domain portion if present. also note we keep the hostname in
# two formats, one in it's unaltered form (at least needed by lustre directory
# parsing) as well as all lc because it displays nicer.
$Host=`hostname`;
chomp $Host;
$Host=(split(/\./, $Host))[0];
$HostLC=lc($Host);
# when was system booted?
$uptime=(split(/\s+/, `cat /proc/uptime`))[0];
$boottime=time-$uptime;
$Distro=cat('/etc/redhat-release') if -e '/etc/redhat-release';
chomp $Distro;
if (-e '/etc/redhat-release')
{
$Distro=cat('/etc/redhat-release');
chomp $Distro;
}
elsif (-e '/etc/SuSE-release')
{
my @temp=split(/\n/, cat('/etc/SuSE-release', 1));
$temp[2]=~/(\d+$)/; # patchlevel
$Distro="$temp[0] SP$1"; # append onto release string as SP
}
elsif (-e '/etc/debian_version')
{
# Both debian and ubuntu have 2 files
$Distro='debian '.cat('/etc/debian_version');
chomp $Distro;
# append distro to base release, if there
if (-e '/etc/lsb-release')
{
my $temp=cat('/etc/lsb-release',1);
$temp=~/DESCRIPTION=(.*)/;
$temp=$1;
$temp=~s/\"//g;
$Distro.=", $temp";
}
}
# For -sD calculations, we need the HZ of the system
$HZ=POSIX::sysconf(&POSIX::_SC_CLK_TCK);
$PageSize=POSIX::sysconf(_SC_PAGESIZE);
# If we have process IO everyone must. This was added in 2.6.23,
# but then only if someone builds the kernel with it enabled, though
# that, will probably change with future kernels.
$processIOFlag=(-e '/proc/self/io') ? 1 : 0;
$slabinfoFlag= (-e '/proc/slabinfo') ? 1 : 0;
$slubinfoFlag= (-e '/sys/slab') ? 1 : 0;
$processCtxFlag=($subsys=~/Z/ && `$Grep ctxt /proc/self/status` ne '') ? 1 : 0;
# just because slab structures there, are they readable? A chunk of extra work, but worth it.
if ($subsys=~/y/i && $slabinfoFlag || $slubinfoFlag)
{
$message='';
$message='/proc/slabinfo' if $slabinfoFlag && !(eval {`cat /proc/slabinfo 2>/dev/null` or die});
$message='/sys/slab' if $slubinfoFlag && !(eval {`cat /proc/slubinfo 2>/dev/null` or die});
if ($message ne '')
{
my $whoami=`whoami`;
chomp $whoami;
disableSubsys('y', "/proc/slabinfo is not readable by $whoami");
$interval=~s/(^\d*):\d+/$1:/ if $subsys!~/z/i; # remove int2 if not needed or we'll get error
}
}
# Get number of ACTIVE CPUs from /proc/stat and in case we're not running on a
# kernel that will set the CPU states (or let us change them), enable CPU flag
$NumCpus=`$Grep cpu /proc/stat | wc -l`;
$NumCpus=~/(\d+)/;
$NumCpus=$1-1;
for (my $i=0; $i<$NumCpus; $i++)
{
$cpuEnabled[$i]=1;
}
$cpusEnabled=$NumCpus;
# Now get the total number the system sees, and if different, reset the
# number as well as a flag for the header
$cpuDisabledFlag=0;
$cpuDisabledMsg='';
$cpusDisabled=0;
if (-e '/sys')
{
my $totalCpus=`ls /sys/devices/system/cpu/|$Grep '^cpu[0-9]'|wc -l`;
chomp $totalCpus;
if ($totalCpus!=$NumCpus)
{
$NumCpus=$totalCpus;
$cpusDisabled++; # for use in header
$cpuDisabledFlag=1;
# we really only have to worry about lower level details of WHO is disabled when
# doing cpu or interrupt stats. Furthermore, if doing cpu, the dynamic processing
# will figure out who's disabled but its too much overhead for interrupts alone so
# we'll do that here one time only. This may have to be do dynamically if a problem.
if ($subsys=~/j/i && $subsys!~/c/i)
{
# CPU0 always online AND no 'online' entry exists!
$cpusEnabled=1;
$cpuEnabled[0]=1;
for (my $i=1; $i<$NumCpus; $i++)
{
my $online=`cat /sys/devices/system/cpu/cpu$i/online`;
chomp $online;
$cpuEnabled[$i]=$online;
$cpusEnabled++ if $online;
$intrptTot[$i]=0;
}
}
}
}
$temp=`$Grep vendor_id /proc/cpuinfo`;
$CpuVendor=($temp=~/: (.*)/) ? $1 : '???';
$temp=`$Grep siblings /proc/cpuinfo`;
$CpuSiblings=($temp=~/: (\d+)/) ? $1 : 1; # if not there assume 1
$temp=`$Grep "cpu cores" /proc/cpuinfo`;
$CpuCores=($temp=~/: (\d+)/) ? $1 : 1; # if not there assume 1
$temp=`$Grep "cpu MHz" /proc/cpuinfo`;
$CpuMHz=($temp=~/: (.*)/) ? $1 : '???';
$Hyper=($CpuSiblings/$CpuCores==2) ? "[HYPER]" : "";
if (-e "/sys/devices/system/node")
{
$CpuNodes=`ls /sys/devices/system/node |$Grep '^node[0-9]'|wc -l`;
}
else
{
# if doesn't exist set nodes to 1 and disable '-sM' if specified
$CpuNodes=1;
disableSubsys('M', "/sys/devices/system/node doesn't exist", 1) if $subsys=~/M/;
}
chomp $CpuNodes;
# /proc read speed test, note various reasons to skip it
# These tests should be updated as we learn more about other distros
$ProcReadTest='no' if $NumCpus<32 || $Kernel lt '2.6.32';
$ProcReadTest='no' if $Distro=~/Red Hat.*release (\S+)/ && $1>=6.2;
$ProcReadTest='no' if $Distro=~/SUSE.*Server (\d+).*SP(\d+)/ && ($1!=11 || $2>=1);
if ($ProcReadTest=~/yes/i)
{
$procReadTested=1; # can call this routine twice!
my $strace=`strace -c cat /proc/stat 2>&1`;
$strace=~/^\s*\S+\s+(\S+).*read$/m;
my $speed=$1;
print "ProcReadSpeed: $speed\n" if $debug * 1;
if ($speed>0.01)
{
# may be going to a lot of effort here but I want to make sure these messages aren't
# recorded as errors and you don't have to include -m to see them as they are pretty
# important to the users to know about this.
my $line1="Slow /proc/stat read speed of $speed seconds";
my $line2="Consider a kernel patch/upgrade. See http://collectl.sourceforge.net/FAQ for more";
my $line3="Change 'ProcReadSpeed' in /etc/collectl.conf to suppress this message in the future";
if ($DaemonFlag)
{
logmsg('W', $line1);
logmsg('I', $line2);
logmsg('I', $line3);
}
else
{
print "$line1\n$line2\n$line3\n";
}
}
}
$Memory=`$Grep MemTotal /proc/meminfo`;
$Memory=(split(/\s+/, $Memory, 2))[1];
chomp $Memory;
$Swap=`$Grep SwapTotal /proc/meminfo`;
$Swap=(split(/\s+/, $Swap, 2))[1];
chomp $Swap;
# B u d d y i n f o
if ($subsys=~/b/i)
{
if (!open BUD, ')
{
$NumBud++
}
close BUD;
}
}
# D i s k C h e c k s
undef @dskOrder;
$dskIndexNext=0;
my @temp=`$Cat /proc/diskstats`;
foreach my $line (@temp)
{
next if $line!~/$DiskFilter/;
my @fields=split(/\s+/, $line);
my $diskName=$fields[3];
$diskName=remapDiskName($diskName) if $diskRemapFlag;
$diskName=~s/cciss\///;
push @dskOrder, $diskName;
$disks{$diskName}=$dskIndexNext++;
}
$dskSeenLast=$dskIndexNext;
logmsg("I", "initDisk initialized $dskIndexNext disks") if $debug & 1;
# I n o d e s
if ($subsys=~/i/)
{
$dentryFlag= (-e '/proc/sys/fs/dentry-state') ? 1 : 0;
$inodeFlag= (-e '/proc/sys/fs/inode-state') ? 1 : 0;
$filenrFlag= (-e '/proc/sys/fs/file-nr') ? 1 : 0;
if ($debug & 1)
{
print "/proc/sys/fs/dentry-state missing\n" if !$dentryFlag;
print "/proc/sys/fs/dentry-state missing\n" if !$inodeFlag;
print "/proc/sys/fs/dentry-state missing\n" if !$filenrFlag;
}
}
# I n t e r c o n n e c t C h e c k s
# Set IB speeds non-conditionally (even if not running IB) and then only for ofed.
# Furthermore assume if mulitple IB interfaces they're all the same speed.
$ibSpeed='??';
if (-e '/sys/class/infiniband')
{
$line=`cat /sys/class/infiniband/*/ports/1/rate 2>&1`;
if ($line=~/\s*(\d+)\s+(\S)/)
{
$ibSpeed=$1;
$ibSpeed*=1000 if $2 eq 'G';
}
}
# if doing interconnect, the first thing to do is see what interconnect
# hardware is present via lspci. Note that from the H/W database, we get
# the following IDS -Quadrics: 14fc, Myricom: 14c1, Mellanox (IB): 15b3
# OR 0c06, QLogic (IB): 1077
# we also have to make sure in the right position of output of lspci command
# so need to be a little clever
$NumXRails=$NumHCAs=0;
$myrinetFlag=$quadricsFlag=$mellanoxFlag=0;
if ($subsys=~/x/i)
{
my $lspciVer=`$Lspci --version`;
$lspciVer=~/ (\d+\.\d+)/;
$lspciVer=$1;
my $lspciVendorField=($lspciVer<2.2) ? 3 : 2;
# Turns out SuSE put 'Class' string back into V2.4.4 without changing
# version number in SLES 10. It also looks like they got it right in
# SLES 11, but who know what will happen in SLES 12!
$lspciVendorField=3 if $Distro=~/SUSE.*10/;
print "lspci -- Version: $lspciVer Vendor Field: $lspciVendorField\n"
if $debug & 1;
$command="$Lspci -n | $Egrep '15b3|0c06|14c1|14fc|1077'";
print "Command: $command\n" if $debug & 1;
@pci=`$command`;
foreach $temp (@pci)
{
# Save the type in case we ever need that level of discrimination.
($vendorID, $type)=split(/:/,(split(/\s+/, $temp))[$lspciVendorField]);
if ($vendorID eq '14c1')
{
printf "WARNING: found myrinet card but no collectl support\n";
}
if ($vendorID eq '14fc')
{
print "Found Quadrics Interconnect\n" if $debug & 2;
$quadricsFlag=1;
elanCheck();
}
if ($vendorID=~/15b3|0c06|1077/)
{
next if $type eq '5a46'; # ignore pci bridge
print "Found Infiniband Interconnect\n" if $debug & 1;
$mellanoxFlag=1;
$HCANames='';
ibCheck('');
}
}
disableSubsys('x', 'no interconnect hardware/drivers found')
if $myrinetFlag+$quadricsFlag+$mellanoxFlag==0;
# User had ability to turn off in case they don't want destructive monitoring
if ($mellanoxFlag)
{
$message='';
$message="Open Fabric IB Stats disabled in $configFile" if -e $SysIB && $PQuery eq '';
$message="Voltaire IB Stats disabled in $configFile" if !-e $SysIB && $PCounter eq '';
if ($message ne '')
{
logmsg("W", $message);
$xFlag=$XFlag=0;
$subsys=~s/x//ig;
$mellanoxFlag=0;
}
if ($mellanoxFlag)
{
# The way forward is clearly OFED
if (-e $SysIB)
{
# This block's job is to make sure perfquery is there
{
print "Looking for 'perfquery' and 'ofed_info'\n" if $debug & 2;
$PQuery=getOfedPath($PQuery, 'perfquery', 'PQuery');
if ($PQuery eq '')
{
disableSubsys('x', "couldn't find perfquery!");
$mellanoxFlag=0;
last;
}
# I hate support questions and this is the place to catch perfquery problems!
# so, if perfquery IS there, since it generates warnings on stderr in V1.5 and
# we don't know the version yet, always ignore them
if ($mellanoxFlag)
{
my $message='';
my $temp=`$PQuery 2>/dev/null`;
$message="Permission denied" if $temp=~/Permission denied/;
$message="Failed to open IB device" if $temp=~/Failed to open/;
$message="Required module missing" if $temp=~/required by/;
$message="No such file or directory" if $temp=~/No such file/;
if ($message ne '')
{
disableSubsys('x', "perfquery error: $message!");
$mellanoxFlag=0;
$PQuery='';
last;
}
}
# perfquery IS there and we can execute it w/o error...
# Can you believe it? PQuery writes its version output to stderr!
$temp=`$PQuery -V 2>&1`;
$temp=~/VERSION: (\d+\.\d+\.\d+)/;
$PQVersion=$1;
}
# perfquery there, but what is ofed's version?
# NOTE - looks like RedHat is no longer shipping ofed
if ($PQuery ne '')
{
if (!-e $OfedInfo)
{
$OfedInfo=getOfedPath($OfedInfo, 'ofed_info', 'OfedInfo');
logmsg('W', "Couldn't find 'ofed_info'. Won't be able to determine OFED version")
if $OfedInfo eq '';
}
# Unfortunately the ofed_info that ships with voltaire adds 5 extra
# line at front end so let's look at first 10 lines for version.
$IBVersion=($OfedInfo ne '' && `$OfedInfo|head -n10`=~/OFED-(.*)/) ? $1 : '???';
print "OFED V: $IBVersion PQ V:$PQVersion\n" if $debug & 2;
}
}
else
{
$IBVersion=(`head -n1 $VoltaireStats`=~/ibstat\s+(.*)/) ? $1 : '???';
print "Voltaire IB V$IBVersion\n" if $debug & 2;
}
}
}
# One last check and this is a doozie! Because we read IB counters by doing
# a read/clear everytime, multiple copies of collectl will step on each other.
# Therefore we can only allow one instance to actually monitor the IB and the
# first one wins, unless we're trying to start a daemon in which case we let
# step on the other [hopefully temporary] instance. Since there are odd cases
# where it may not always catch exception, one can override checking in .conf
if ($IbDupCheckFlag && $mellanoxFlag)
{
my $myppid=getppid();
$command="$Ps axo pid,cmd | $Grep collectl | $Grep -vE 'grep|ssh'";
foreach my $line (`$command`)
{
$line=~s/^\s+//; # some pids have leading white space
my ($pid, $procCmd)=split(/ /, $line, 2);
next if $pid==$$ || $pid==$myppid; # check ppid in case started by a script
# There are just too many ways one can specify the subsystems whether it's
# overriding the DaemonCommands or SubsysCore in collectl.conf, using an
# alternate collectl.conf or specifying --subsys instead of -s and I'm
# just not going to go there [for now] as it's complicated enough, hence
# '$IbDupCheckFlag'
# If not running as a daemon, '$procCmd' has the command invocation string
# from the 'ps' above. If a daemon, we need to pull it out of collectl.cont.
my $tempDaemonFlag=($procCmd=~/-D/) ? 1 : 0;
if ($tempDaemonFlag)
{
# This is getting even uglier, but if someone chose to duplicate
# 'DaemonCommands' and comment one out, we really need to look for
# the last uncommented one.
foreach my $cmd (`$Grep 'DaemonCommands =' $configFile`)
{
next if $cmd=~/^#/;
$procCmd=$cmd;
}
}
# Now that we have the full command passed to collectl, pull out -s (if any)
# which may be surrounded by optional white space.
chomp $procCmd;
$procSubsys=($procCmd=~/-s\s*(\S+)\s*/) ? $1 : '';
# The default subsys is different for daemon and interactive use
# if no -s, we use default and if there, assume we're overriding
$tempSubsys=($tempDaemonFlag) ? $SubsysDefDaemon : $SubsysDefInt;
# So now we need to figure out what actual subsystems are in use
# by that instance in case it was started with either +/- OR
# a fixed set
if ($procSubsys=~/^[\+\-]/)
{
# the stolen from main collectl switch validation code noting
# we don't need to validate the switches since done when
# daemon started
if ($procSubsys=~/-(.*)/)
{
my $pat=$1;
$pat=~s/\+.*//; # if followed by '+' string
$tempSubsys=~s/[$pat]//g;
}
if ($procSubsys=~/\+(.*)/)
{
my $pat=$1;
$pat=~s/-.*//; # if followed by '-' string
$tempSubsys.=$pat;
}
}
elsif ($procSubsys ne '')
{
$tempSubsys=$procSubsys;
}
# At this point if there IS an instance of collectl running with -sx,
# we need to disable it here, unless we're a daemon in which case we
# just log a warning.
if ($tempSubsys=~/x/i)
{
if (!$daemonFlag)
{
disableSubsys('x', 'another instance already monitoring Infiniband');
}
else
{
logmsg("W", "another instance is monitoring IB and the stats will be in error until it is stopped");
}
last;
}
}
}
}
# Let's always get the platform name if dmidecode is there
if ($Dmidecode ne '')
{
$ProductName=($rootFlag) ? `$Dmidecode | grep -m1 'Product Name'` : '';
$ProductName=~s/\s*Product Name: //;
chomp $ProductName;
$ProductName=~s/\s*$//; # some have trailing whitespace
}
# E n v i r o n m e n t a l C h e c k s
if ($subsys=~/E/ && $envTestFile eq '')
{
# Note that these tests are in the reverse order since the last value of $message
# in the one reported AND only if not using a 'test' file for data source.
my $message='';
$message="'IpmiCache' not defined or specifies a directory"
if $IpmiCache eq '' || -d $IpmiCache;
$message="cannot find /dev/ipmi* (is impi_si loaded?)"
if !-e '/dev/ipmi0' && !-e '/dev/ipmi/0' && !-e '/dev/ipmidev/0';
$message="cannot find 'ipmitool' in '$ipmitoolPath'"
if $Ipmitool eq '';
$message="you must be 'root' to do environmental monitoring"
if !$rootFlag;
if ($message eq '')
{
# If specified by --envopts, set -d for ipmitool
$Ipmitool.=" -d $1" if $envOpts=~/(\d+)/;
logmsg('I', "Initialized ipmitool cache file '$IpmiCache'");
my $command="$Ipmitool sdr dump $IpmiCache";
# If we can't dump the cache, something is wrong so make sure we pass along
# error and disable E monitoring. Ok to create 'exec' below since we'll
# never execute it
$message=`$command 2>&1`;
if ($message=~/^Dumping/)
{
# Create 'exec' option file in save directory as cache, but only for
# those options that actually return data
my $cacheDir=dirname($IpmiCache);
$ipmiExec="$cacheDir/collectl-ipmiexec";
if (open EXEC, ">$ipmiExec")
{
$message=''; # indicates no errors for test below
foreach my $type (split(/,/, $IpmiTypes))
{
my $command="$Ipmitool -S $IpmiCache sdr type $type";
next if `$command` eq '';
print EXEC "sdr type $type\n";
}
close EXEC;
}
else
{
$message="couldn't create '$ipmiExec'";
}
}
}
disableSubsys('E', $message) if $message ne '';
}
# find all the networks and when possible include their speeds
undef @temp;
$netIndexNext=0;
$NetWidth=$netOptsW; # Minimum size
$null=($debug & 1) ? '' : '2>/dev/null';
my $interval1=(split(/:/, $interval))[0];
# but first look up all the network speed in /sys/devices and load them into a hash
# for easier access in the loop below
my $command="find /sys/devices/ 2>&1 | grep net | grep speed";
open FIND, "$command|" or logmsg('E', "couldn't execute '$command'");
while (my $line=)
{
chomp $line;
my $mode=$line;
$mode=~s/speed/operstate/;
$mode=`cat $mode`;
next if $mode!~/up|unknown/;
my $speed=`cat $line 2>&1`;
chomp $speed;
$line=~/.*\/(\S+)\/speed/;
my $netName=$1;
$speed='??' if $netName=~/^vnet/; # hardcoded in kernel to 10 which causes bogus msgs later
$netSpeeds{$netName}=$speed if $speed!~/Invalid/; # this can happen on a VM where operstate IS up
#print "set netSpeeds{$netName}=$speed\n";
}
close FIND;
# Since this routine can get called multiple times during
# initialization, we need to make sure @netOrder gets clean start.
undef @netOrder;
@temp=`$Grep -v -E "Inter|face" /proc/net/dev`;
foreach my $temp (@temp)
{
next if $rawNetFilter ne '' && $temp!~/$rawNetFilter/;
$temp=~/^\s*(\S+)/; # most names have leading whitespace
$netName=$1;
$netName=~s/:.*//; # get rid of : AND possible stats if no whitespace
$NetWidth=length($netName) if length($netName)>$NetWidth;
$speed=($netName=~/^ib/) ? $ibSpeed : $netSpeeds{$netName};
$speed='??' if !defined($speed);
push @netOrder, $netName;
$netIndex=$netIndexNext;
$networks{$netName}=$netIndexNext++;
# Since speeds are in Mb we really need to multiple by 125 to conver to KB
$NetMaxTraffic[$netIndex]=($speed ne '' && $speed ne '??') ?
2*$interval1*$speed*125 : 2*$interval1*$DefNetSpeed*125;
}
$netSeenLast=$netIndexNext;
$NetWidth++; # make room for trailing colon
# S C S I C h e c k s
# not entirely sure what to do with SCSI info, but if feels like a good
# thing to have. also, if no scsi present deal accordingly
undef @temp;
$ScsiInfo='';
if (-e "/proc/scsi/scsi")
{
@temp=`$Grep -E "Host|Type" /proc/scsi/scsi`;
foreach $temp (@temp)
{
if ($temp=~/^Host: scsi(\d+) Channel: (\d+) Id: (\d+) Lun: (\d+)/)
{
$scsiHost=$1;
$channel=$2;
$id=$3;
$lun=$4;
}
if ($temp=~/Type:\s+(\S+)/)
{
$scsiType=$1;
$type="??";
$type="SC" if $scsiType=~/scanner/i;
$type="DA" if $scsiType=~/Direct-Access/i;
$type="SA" if $scsiType=~/Sequential-Access/i;
$type="CD" if $scsiType=~/CD-ROM/i;
$type="PR" if $scsiType=~/Processor/i;
$ScsiInfo.="$type:$scsiHost:$channel:$id:$lun ";
}
}
$ScsiInfo=~s/ $//;
}
# L u s t r e C h e c k s
$CltFlag=$MdsFlag=$OstFlag=0;
$NumLustreFS=$numBrwBuckets=0;
if ($subsys=~/l/i)
{
if ((`ls /lib/modules/*/kernel/net/lustre 2>/dev/null|wc -l`==0) &&
(`ls /lib/modules/*/*/kernel/net/lustre 2>/dev/null|wc -l`==0))
{
disableSubsys('l', 'this system does not have lustre modules installed');
}
else
{
# Get Luster and SFS Versions before looking at any data structions in the
# 'lustreCheck' routines because things change over time
$temp=`$Lctl lustre_build_version 2>/dev/null`;
$temp=~/version: (.+?)-/m;
$cfsVersion=$1;
$sfsVersion='';
if (-e '/etc/sfs-release')
{
$temp=cat('/etc/sfs-release');
$temp=~/(\d.*)/;
$sfsVersion=$1;
}
elsif (-e "/usr/sbin/sfsmount" && -e $Rpm)
{
# XC and client enabler
$llite=`$Rpm -qa | $Grep lustre-client`;
$llite=~/lustre-client-(.*)/;
$sfsVersion=$1;
}
$OstWidth=$FSWidth=0;
$NumMds=$NumOst=0;
$MdsNames=$OstNames=$lustreCltInfo='';
$inactiveOstFlag=0;
lustreCheckClt();
lustreCheckMds();
lustreCheckOst();
print "Lustre -- CltFlag: $CltFlag NumMds: $NumMds NumOst: $NumOst\n"
if $debug & 8;
disableSubsys('l', "no lustre services running and I don't know its type. You will need to use --lustsvc to force type.")
if $CltFlag+$NumMds+$NumOst==0 && $lustreSvcs eq '';
# Global to count how many buckets there are for brw_stats
@brwBuckets=(1,2,4,8,16,32,64,128,256);
push @brwBuckets, (512,1024) if $sfsVersion ge '2.2';
$numBrwBuckets=scalar(@brwBuckets);
# if we're doing lustre DISK stats, figure out what kinds of disks
# and then build up a list of them for collection to use. To keep switch
# error processing clean, only try to open the file if an MDS or OSS.
# Since services may not be up, we also need to look at '$lustreSvcs',
# though ultimately we'll only set the disk types and the maximum buckets
if ($subsys=~/l/i && $lustOpts=~/D/ && ($MdsFlag || $OstFlag || $lustreSvcs=~/[mo]/i))
{
# The first step is to build up a hash of the sizes of all the
# existing partitions. Since we're only doing this once, a 'cat's
# overhead should be minimal
@partitions=`cat /proc/partitions`;
foreach $part (@partitions)
{
# ignore blank lines and header
next if $part=~/^\s*$|^major/;
# now for the magic. Get the partition size and name, but ignore
# cciss devices on controller 0 OR any devices with partitions
# noting cciss device partitions end in 'p-digit' and sd partitions
# always end in a digit.
($size, $name)=(split(/\s+/, $part))[3,4];
$name=~s/cciss\///;
next if $name=~/^c0|^c.*p\d$|^sd.*\d$/;
$partitionSize{$name}=$size;
}
# Determine which directory to look in based on whether or not there
# is an EVA present. If so, we look at 'sd' stats; otherwize 'cciss'
$LusDiskNames='';
$LusDiskDir=(-e '/proc/scsi/sd_iostats') ?
'/proc/scsi/sd_iostats' : '/proc/driver/cciss/cciss_iostats';
# Now find all the stat files, noting that in the case of cciss, we
# always skip c0 disks since they're local ones... Also note that
# if we're doing a showHeader with -Lm or -Lo on a client, the file
# isn't there AND we don't want to report an error either.
$openFlag=(opendir(DIR, $LusDiskDir)) ? 1 : 0;
logmsg('F', "Disk stats requested but couldn't open '$LusDiskDir'")
if !$openFlag && !$showHeaderFlag;
while ($diskname=readdir(DIR))
{
next if $diskname=~/^\.|^c0/;
# if this has a partition within the range of a service lun,
# ignore it.
if ($partitionSize{$diskname}/(1024*1024)<$LustreSvcLunMax)
{
print "Ignoring $diskname because its size of ".
"$partitionSize{$diskname} is less than ${LustreSvcLunMax}GB\n"
if $debug & 1;
next;
}
push @LusDiskNames, $diskname;
$LusDiskNames.="$diskname ";
}
$LusDiskNames=~s/ $//;
$NumLusDisks=scalar(@LusDiskNames);
$LusMaxIndex=($LusDiskNames=~/sd/) ? 16 : 24;
}
}
}
# S L A B C h e c k s
# Header for /proc/slabinfo changed in 2.6
if ($slabinfoFlag && $subsys=~/y/i)
{
$SlabGetProc=($slabFilt eq '') ? 99 : 14;
$temp=`head -n 1 /proc/slabinfo`;
$temp=~/(\d+\.\d+)/;
$SlabVersion=$1;
$NumSlabs=`cat /proc/slabinfo | wc -l`*1;
chomp $NumSlabs;
$NumSlabs-=2;
if ($SlabVersion!~/^1\.1|^2/)
{
# since 'W' will echo on terminal, we only use when writing to files
$severity=(defined($opt_s)) ? "E" : "I";
$severity="W" if $logToFileFlag;
logmsg($severity, "unsupported /proc/slabinfo version: $SlabVersion");
$subsys=~s/y//gi;
$yFlag=$YFlag=0;
}
}
}
# Why is initFormat() so damn big?
#
# Since logs can be analyzed on a system on which they were not generated
# and to avoid having to read the actual data to determine things like how
# many cpus or disks there are, this info is written into the log file
# header. initFormat() then reads this out of the head and initialized the
# corresponding variables.
#
# Counters are always incrementing (until they wrap) and therefore to get the
# value for the current interval one needs decrement it by the sample from
# the previous interval. Therefore, theere are 3 different types of
# variables to deal with:
# - current sample: some 'root', ends in 'Now'
# - last sample: some 'root', end in 'Last'
# - true value: 'root' only - rootNow-rootLast
#
# To make all this work the very first time through, all 'Last' variables
# need to be initialized to 0 both to suppress -w initialization warnings AND
# because it's good coding practice. Furthermore, life is a lot cleaner just
# to initialize everything whether we've selected the corresponding subsystem
# or not. Furthermore, since it is possible to select a subsystem in plot
# mode for which we never gathered any data, we need to initialize all the
# printable values to 0s as well. That's why there is so much crap in
# initFormat().
sub initFormat
{
my $playfile=shift;
my ($day, $mon, $year, $i, $recsys, $host);
my ($version, $datestamp, $timestamp, $interval);
$temp=(defined($playfile)) ? $playfile : '';
print "initFormat($temp)\n" if $debug & 1;
# Constants local to formatting
$OneKB=1024;
$OneMB=1024*1024;
$OneGB=1024*1024*1024;
$TenGB=$OneGB*10;
# in normal mode we report "/sec", but with -on we report "/int", noting
# this is also appended to plot format headers
$rate=$options!~/n/ ? "/sec" : "/int";
if (defined($playfile))
{
$header=getHeader($playfile);
return undef if $header eq '';
# save the first two lines of the header for writing into the new header.
# since the Deamon Options have been renamed in V1.5.3 we need to get a
# little trickier to handle both. Since they are so specific I'm leaving
# them global.
$header=~/(Collectl.*)/;
$recHdr1=$1;
$recHdr2=(($header=~/(Daemon Options: )(.*)/ || $header=~/(DaemonOpts: )(.*)/) && $2 ne '') ? "$1$2" : "";
$header=~/Collectl:\s+V(\S+)/;
$version=$1;
$hiResFlag=$1 if $header=~/HiRes:\s+(\d+)/; # only after V1.5.3
$boottime=($header=~/Booted:\s+(\S+)/) ? $1 : 0;
$Distro='';
if ($header=~/Distro:\s+(.+)/) # was optional before 'Platform' added
{
$Distro=$1;
$ProductName=$1 if $Distro=~s/Platform: (.*)//;
}
# Prior to collect V3.2.1-4, use the header to determine the type of nfs data in the
# file noting very old versions used SubOpts.
$recNfsFilt=$1 if $header=~/NfsFilt: (\S*) \S/;
$subOpts=($header=~/SubOpts:\s+(\S*)\s*Options/) ? $1 : ''; # pre V3.2.1-4
if ($version lt '3.2.1-4')
{
$nfsOpts=($header=~/NfsOpts: (\S*)\s*Interval/) ? $1 : $subOpts;
$nfsOpts=~s/[BDMORcom]//g; # in case it came from SubOpts remove lustre stuff
if ($version lt '3.2.1-3')
{
$recNfsFilt=($nfsOpts=~/C/) ? 'c' : 's';
$recNfsFilt.=($nfsOpts=~/([234])/) ? $1 : 3;
}
else
{
# very limited release
$recNfsFilt=($nfsOpts=~/C/) ? 'c3,c4' : 's3,s4';
}
}
if ($header=~/TcpFilt:\s+(\S+)/)
{
# remember, even if an option is not recorded we still report on it
my $recOpts=(defined($1)) ? $1 : $tcpFiltDefault;
$tcpFilt=$recOpts if $tcpFilt eq '';
}
# Users CAN overrider LustOpts so we need to do it this way, again accounting for
# older versions of collectl storing them as part of SubOpts
if ($lustOpts eq '')
{
$lustOpts=($header=~/LustOpts: (\S*)\s*Services/) ? $1 : $subOpts;
$lustOpts=~s/[23C]//g; # remove nfs options
}
# we want to preserve original subsys from the header, but we
# also want to override it if user did a -s. If user specified a
# +/- we also need to deal with as in collectl.pl, but in this
# case without the error checking since it already passed through.
# NOTE - rare, but if not subsys, set to ' ' also noting '' won't work
# in regx in collectl after call to this routine
$header=~/SubSys:\s+(\S*) /;
$recSubsys=$subsys=($1!~/Options/) ? $1 : ' ';
$recHdr1.=" Subsys: $subsys";
$recSubsys=$subsys='Y' if $topSlabFlag && $userSubsys eq '';
$recSubsys=$subsys='Z' if $topProcFlag && $userSubsys eq '';
# reset subsys based on what was recorded and -s
$subsys=mergeSubsys($recSubsys);
$subsys.='Y' if $subsys!~/Y/ && $topSlabFlag; # if --top need to include Y or Z if not in -s
$subsys.='Z' if $subsys!~/Z/ && $topProcFlag;
# I'm not sure the Mds/Ost/Clt names still need to be initialized
# but it can't hurt. Clearly the 'lustre' variables do.
$MdsNames=$OstNames=$lustreClts='';
$lustreMdss=$lustreOsts=$lustreClts='';
# This can only happen with pre 3.0.0 version of collectl
if ($subsys=~/LL/)
{
$subsys=~s//L/;
$lustOpts.='O';
}
# We ONLY override the settings for the raw file, never any others.
# Even though currently only 'rawp' files, we're doing pattern match below
# with [p] to make easier to add others if we ever need to.
$playfile=~/(.*-\d{8})-\d{6}\.raw([p]*)/;
if (defined($playbackSettings{$1}) && $2 eq '')
{
# NOTE - when -L not specified for lustre, $lustreSvcs will end up being
# set to the combined values of all files for this prefix
($subsys, $lustreSvcs, $lustreMdss, $lustreOsts, $lustreClts)=
split(/\|/, $playbackSettings{$1});
print "OVERRIDES - Subsys: $subsys LustreSvc: $lustreSvcs ".
"MDSs: $lustreMdss Osts: $lustreOsts Clts: $lustreClts\n"
if $debug & 2048;
}
print "Playfile: $playfile Subsys: $subsys\n" if $debug & 1;
setFlags($subsys);
# In case not in current file header but defined within set for prefix/date
$CltFlag=$MdsFlag=$OstFlag=$NumMds=$NumOst=$OstWidth=$FSWidth=0;
$MdsNames=$lustreMdss if $lustreMdss ne '';
$OstNames=$lustreOsts if $lustreOsts ne '';
# Maybe some day we can get rid of pre 1.5.0 support?
$numBrwBuckets=0;
if ($header=~/Lustre/ && $version ge '1.5.0')
{
# Remember, we could have cfs without sfs so need 2 separate pattern tests
$cfsVersion=$sfsVersion='';
if ($version ge '2.1')
{
$header=~/CfsVersion:\s+(\S+)/;
$cfsVersion=$1;
$header=~/SfsVersion:\s+(\S+)/;
$sfsVersion=$1;
}
# In case not already defined (for single or consistent files, these are
# not specified as overrides), get them from the file header. Note that
# when no osts, this will grab the next line it I include \s* after
# OstNames:, so for now I'm doing it this way and chopping leading space.
$MdsHdrNames=$OstHdrNames='';
if ($header=~/MdsNames:\s+(.*)\s*NumOst:\s+\d+\s+OstNames:(.*)$/m)
{
$MdsHdrNames=$1;
$OstHdrNames=$2;
$OstHdrNames=~s/\s+//;
$MdsNames=($lustreMdss ne '') ? $lustreMdss : $MdsHdrNames;
$OstNames=($lustreOsts ne '') ? $lustreOsts : $OstHdrNames;
}
if ($MdsNames ne '')
{
@MdsMap=remapLustreNames($MdsHdrNames, $MdsNames, 0) if $MdsHdrNames ne '';
foreach $name (split(/ /, $MdsNames))
{
$NumMds++;
$MdsFlag=1;
}
}
if ($OstNames ne '')
{
# This build list for interpretting input from 'raw' file if there is any
@OstMap=remapLustreNames($OstHdrNames, $OstNames, 0) if $OstHdrNames ne '';
# This builds data needed for display
foreach $name (split(/ /, $OstNames))
{
$lustreOstName[$NumOst]=$name;
$lustreOsts[$NumOst++]=$name;
$OstWidth=length($name) if length($name)>$OstWidth;
$OstFlag=1;
}
}
if ($header=~/CltInfo:\s+(.*)$/m)
{
$CltHdrNames=$1;
$lustreCltInfo=($lustreCltInfo ne '') ? $lustreCltInfo : $CltHdrNames;
}
undef %fsNames;
$CltFlag=$NumLustreFS=$NumLustreCltOsts=0;
$lustreCltInfo=$lustreClts if $lustreClts ne '';
if ($lustreCltInfo ne "")
{
$CltFlag=1;
foreach $name (split(/ /, $lustreCltInfo))
{
($fsName, $ostName)=split(/:/, $name);
$lustreCltFS[$NumLustreFS++]=$fsName if !defined($fsNames{$fsName});
$fsNames{$fsName}=1;
$FSWidth=length($fsName) if length($fsName)>$FSWidth;
# if osts defined, we just overwrite anything with did for the non-ost
if ($ostName ne '')
{
$lustreCltOsts[$NumLustreCltOsts]=$ostName;
$lustreCltOstFS[$NumLustreCltOsts]=$fsName;
$OstWidth=length($ostName) if length($ostName)>$OstWidth;
$NumLustreCltOsts++;
}
}
@CltFSMap= remapLustreNames($CltHdrNames, $lustreCltInfo, 1)
if defined($CltHdrNames);
@CltOstMap=remapLustreNames($CltHdrNames, $lustreCltInfo, 2)
if defined($CltHdrNames);
}
print "CLT: $CltFlag OST: $OstFlag MDS: $MdsFlag\n" if $debug & 1;
# if disk I/O stats specified in header, init appropriate variables
if ($header=~/LustreDisks.*Names:\s+(.*)/)
{
@lusDiskDirs=split(/\s+/, $1);
$NumLusDisks=scalar(@lusDiskDirs);
$LusDiskNames=$1;
@LusDiskNames=split(/\s+/, $LusDiskNames);
}
}
else # PRE 1.5.0 lustre stuff goes here...
{
if ($header=~/NumOsts:\s+(\d+)\s+NumMds:\s+(\d+)/)
{
$NumOst=$1;
$NumMds=$2;
$OstNames=$MdsNames='';
for ($i=0; $i<$NumOst; $i++)
{
$OstMap[$i]=$i;
$OstNames.="Ost$i ";
$lustreOsts[$i]="Ost$i";
$OstWidth=length("Ost$i") if length("ost$i")>$OstWidth;
$OstFlag=1;
}
$OstNames=~s/ $//;
for ($i=0; $i<$NumMds; $i++)
{
$MdsMap[$i]=$i;
$MdsNames.="Mds$i ";
$MdsFlag=1;
}
$MdsNames=~s/ $//;
}
$NumLustreFS=$NumLustreCltOsts=0;
if ($header=~/FS:\s+(.*)\s+Luns:\s+(.*)\s+LunNames:\s+(.*)$/m)
{
$CltFlag=1;
$tempFS=$1;
$tempLuns=$2;
$tempFSNames=$3;
foreach $fsName (split(/ /, $tempFS))
{
$CltFSMap[$NumLustreFS]=$NumLustreFS;
$lustreCltFS[$NumLustreFS]=$fsName;
$FSWidth=length($fsName) if length($fsName)>$FSWidth;
$NumLustreFS++;
}
# If defined, user did --lustopts O and need to reset FS info
# Also note that since these numbers appear in raw data, we can't use a
# simple index but rather need lun number
if ($tempLuns ne '')
{
# The lun numbers will be mapped into OSTs
foreach $lunNum (split(/ /, $tempLuns))
{
$CltFSMap[$lunNum]=$NumLustreCltOsts;
$CltOstMap[$lunNum]=$NumLustreCltOsts;
$lustreCltOsts[$NumLustreCltOsts]=$lunNum;
$OstWidth=length($lunNum) if length($lunNum)>$FSWidth;
$NumLustreCltOsts++;
}
$NumLustreFS=0;
foreach $fsName (split(/ /, $tempFSNames))
{
$lustreCltOstFS[$NumLustreFS]=$fsName;
$FSWidth=length($fsName) if length($fsName)>$FSWidth;
$NumLustreFS++;
}
}
}
}
$header=~/Host:\s+(\S+)/;
$Host=$1;
$HostLC=lc($Host);
# we need this for timezone conversions...
$header=~/Date:\s+(\d+)-(\d+)/;
$datestamp=$1;
$timestamp=$2;
$timesecs=$timezone=''; # for logs generated with older versions
if ($header=~/Secs:\s+(\d+)\s+TZ:\s+(.*)/)
{
$timesecs=$1;
$timezone=$2;
}
# Allows us to move its location in the header
$header=~/Interval: (\S+)/;
$interval=$1;
# For -s p calculations, we need the HZ of the system
$header=~/HZ:\s+(\d+)\s+Arch:\s+(\S+)/;
$HZ=$1;
$SrcArch=$2;
# In case pagesize not defined in header (for earlier versions
# of collectl) pick a default based on architecture;
$PageSize=($SrcArch=~/ia64/) ? 16384 : 4096;
$PageSize=$1 if $header=~/PageSize:\s+(\d+)/;
# Even though we don't do anything with CPU, Speed, Cores and Siblings we need
# to put them in new header.
$header=~/Cpu:\s+(.*) Speed/;
$CpuVendor=$1;
$header=~/Speed\(MHz\): (\S+)/;
$CpuMHz=$1;
$header=~/Cores: (\d+)/;
$CpuCores=$1;
$header=~/Siblings: (\d+)/;
$CpuSiblings=$1;
$header=~/Nodes: (\d+)/;
$CpuNodes=$1;
# when playing back from a file we need to make sure the KERNEL is that of
# the file and not the one the data was collected on.
$header=~/Kernel:\s+(\S+)/;
$Kernel=$1;
error("collectl no longer supports 2.4 kernels") if $Kernel=~/^2\.4/;
$header=~/NumCPUs:\s+(\d+)/;
$NumCpus=$1;
$Hyper=($header=~/HYPER/) ? "[HYPER]" : "";
$header=~/NumBud:\s+(\d+)/;
$NumBud=$1;
$flags=($header=~/Flags:\s+(\S+)/) ? $1 : '';
$tworawFlag= ($flags=~/[g2]/) ? 1 : 0;
$processIOFlag= ($flags=~/i/) ? 1 : 0;
$slubinfoFlag= ($flags=~/s/) ? 1 : 0;
$processCtxFlag= ($flags=~/x/) ? 1 : 0;
$cpuDisabledFlag=($flags=~/D/) ? 1 : 0;
# If we're not processing CPU data, this message will never be set so
# just initialized for all cases.
$cpuDisabledMsg='';
$header=~/Memory:\s+(\d+)/;
$Memory=$1;
# Since disks are discovered dynamically all we need to init a few pointers.
$dskIndex=$dskSeenLast=0;
# networks are dynamic too but also messier because while we can't get speeds in playback mode
# we do need those that have been recorded so our 'bogus' checks.
$header=~/NumNets:\s+(\d+)\s+NetNames:\s+(.*)/;
$numNets=$1;
$netNames=$2;
$NetWidth=$netOptsW;
my $netIndex=0;
my $interval1=(split(/:/, $interval))[0];
foreach my $netName (split(/ /, $netNames))
{
my $speed=($netName=~/:(\d+)/) ? $1 : $DefNetSpeed;
$netName=~s/(\S+):.*/$1/;
$NetMaxTraffic[$netIndex]=2*$interval1*$speed*125;
$NetWidth=length($netName) if $NetWidth2;
}
# Now get OFED/Perqquery versions which for earlier versions were not in header
# Not clear if we really need these in playback mode but since we may some day...
$IBVersion=($header=~/IBVersion:\s+(\S+)/) ? $1 : '';
$PQVersion=($header=~/PQVersion:\s+(\S+)/) ? $1 : '';
}
# Scsi info is optional
$ScsiInfo=($header=~/SCSI:\s+(.*)/) ? $1 : '';
# Pass header to import routines BUT only if they have a callback defined
for (my $i=0; $i<$impNumMods; $i++)
{ &{$impGetHeader[$i]}(\$header) if defined(&{$impGetHeader[$i]});}
}
# Initialize global arrays with sizes of buckets for lustre brw stats and
# not to worry if lustre not there.
@brwBuckets=(1,2,4,8,16,32,64,128,256);
push @brwBuckets, (512,1024) if defined($sfsVersion) && $sfsVersion ge '2.2';
$numBrwBuckets=scalar(@brwBuckets);
# same thing for lustre disk state though these are a little tricker.
if ($LusDiskNames=~/sd/)
{
@diskBuckets=(.5,1,2,4,8,16,32,64,128,256,512,1024,2048,4096,8192,16384);
}
else
{
@diskBuckets=(.5,1,2,4,8,16,32,63,64,65,80,96,112,124,128,129,144,252,255,256,257,512,1024,2048);
}
$LusMaxIndex=scalar(@diskBuckets);
# this inits lustre variables in both playback and collection modes.
initLustre('o', 0, $NumOst);
initLustre('m', 0, $NumMds);
initLustre('c', 0, $NumLustreFS);
initLustre('c2', 0, $NumLustreCltOsts) if $NumLustreCltOsts ne '-';
# I n i t ' C o r e ' V a r i a b l e s
# when we're generating plot data and we're either not collecting
# everything or we're in playback mode and it's not all in raw file, make
# sure all the core variables that get printed have been initialized to 0s.
# for disks, nets and pars the core variables are the totals and so get
# initialized in the initInterval() routine every cycle
$i=$NumCpus;
$userP[$i]=$niceP[$i]=$sysP[$i]=$idleP[$i]=$totlP[$i]=0;
$irqP[$i]=$softP[$i]=$stealP[$i]=$waitP[$i]=0;
for (my $i=0; $i<$CpuNodes; $i++)
{
foreach my $numa ('used', 'free', 'slab', 'map', 'anon', 'lock', 'inact',)
{ $numaMem[$i]->{$numa}=$numaMem[$i]->{$numa.'C'}=0; }
foreach my $hits ('for', 'miss', 'hits')
{ $numaStat[$i]->{$hits}=0; }
}
$dentryNum=$dentryUnused=$filesAlloc=$filesMax=$inodeUsed=$inodeMax=0;
$loadAvg1=$loadAvg5=$loadAvg15=$loadRun=$loadQue=$ctxt=$intrpt=$proc=0;
$memDirty=$clean=$target=$laundry=$memAct=$memInact=0;
$procsRun=$procsBlock=0;
$pagein=$pageout=$swapin=$swapout=$swapTotal=$swapUsed=$swapFree=0;
$pagefault=$pagemajfault=0;
$memTot=$memUsed=$memFree=$memShared=$memBuf=$memCached=$memSlab=$memAnon=$memMap=$memCommit=$memLocked=0;
$memHugeTot=$memHugeFree=$memHugeRsvd=$memSUnreclaim=0;
$sockUsed=$sockTcp=$sockOrphan=$sockTw=$sockAlloc=0;
$sockMem=$sockUdp=$sockRaw=$sockFrag=$sockFragM=0;
# extended memory stats, just in case some are missing
$pageFree=$pageActivate=0;
$pageAllocDma=$pageAllocDma32=$pageAllocNormal=$pageAllocMove=0;
$pageRefillDma=$pageRefillDma32=$pageRefillNormal=$pageRefillMove=0;
$pageStealDma=$pageStealDma32=$pageStealNormal=$pageStealMove=0;
$pageKSwapDma=$pageKSwapDma32=$pageKSwapNormal=$pageKSwapMove=0;
$pageDirectDma=$pageDirectDma32=$pageDirectNormal=$pageDirectMove=0;
# Lustre MDS stuff - in case no data
$lustreMdsReintCreate=$lustreMdsReintLink=$lustreMdsReintSetattr=0;
$lustreMdsReintRename=$lustreMdsReintUnlink=$lustreMdsReint=0;
$lustreMdsGetattr=$lustreMdsGetattrLock=$lustreMdsStatfs=0;
$lustreMdsGetxattr=$lustreMdsSetxattr=$lustreMdsSync=0;
$lustreMdsConnect=$lustreMdsDisconnect=0;
# Common nfs stats
$rpcCCalls=$rpcSCalls=$rpcBadAuth=$rpcBadClnt=$rpcRetrans=$rpcCredRef=0;
$nfsPkts=$nfsUdp=$nfsTcp=$nfsTcpConn=0;
# V2
$nfs2CNull=$nfs2CGetattr=$nfs2CSetattr=$nfs2CRoot=$nfs2CLookup=$nfs2CReadlink=
$nfs2CRead=$nfs2CWrcache=$nfs2CWrite=$nfs2CCreate=$nfs2CRemove=$nfs2CRename=
$nfs2CLink=$nfs2CSymlink=$nfs2CMkdir=$nfs2CRmdir=$nfs2CReaddir=$nfs2CFsstat=$nfs2CMeta=0;
$nfs2SNull=$nfs2SGetattr=$nfs2SSetattr=$nfs2SRoot=$nfs2SLookup=$nfs2SReadlink=
$nfs2SRead=$nfs2SWrcache=$nfs2SWrite=$nfs2SCreate=$nfs2SRemove=$nfs2SRename=
$nfs2SLink=$nfs2SSymlink=$nfs2SMkdir=$nfs2SRmdir=$nfs2SReaddir=$nfs2SFsstat=$nfs2SMeta=0;
# V3
$nfs3CNull=$nfs3CGetattr=$nfs3CSetattr=$nfs3CLookup=$nfs3CAccess=$nfs3CReadlink=0;
$nfs3CRead=$nfs3CWrite=$nfs3CCreate=$nfs3CMkdir=$nfs3CSymlink=$nfs3CMknod=$nfs3CRemove=0;
$nfs3CRmdir=$nfs3CRename=$nfs3CLink=$nfs3CReaddir=$nfs3CReaddirplus=$nfs3CFsstat=0;
$nfs3CFsinfo=$nfs3CPathconf=$nfs3CCommit=$nfs3CMeta=0;
$nfs3SNull=$nfs3SGetattr=$nfs3SSetattr=$nfs3SLookup=$nfs3SAccess=$nfs3SReadlink=0;
$nfs3SRead=$nfs3SWrite=$nfs3SCreate=$nfs3SMkdir=$nfs3SSymlink=$nfs3SMknod=$nfs3SRemove=0;
$nfs3SRmdir=$nfs3SRename=$nfs3SLink=$nfs3SReaddir=$nfs3SReaddirplus=$nfs3SFsstat=0;
$nfs3SFsinfo=$nfs3SPathconf=$nfs3SCommit=$nfs3SMeta=0;
# V4
$nfs4CNull=$nfs4CRead=$nfs4CWrite=$nfs4CCommit=$nfs4CSetattr=$nfs4CFsinfo=0;
$nfs4CAccess=$nfs4CGetattr=$nfs4CLookup=$nfs4CRemove=$nfs4CRename=$nfs4CLink=0;
$nfs4CSymlink=$nfs4CCreate=$nfs4CPathconf=$nfs4CReadlink=$nfs4CReaddir=$nfs4CMeta=0;
$nfs4SAccess=$nfs4SCommit=$nfs4SCreate=$nfs4SGetattr=$nfs4SLink=$nfs4SLookup=0;
$nfs4SRead=$nfs4SReaddir=$nfs4SReadlink=$nfs4SRemove=$nfs4SRename=$nfs4SSetattr=0;
$nfs4SWrite=$nfs4SMeta=0;
# tcp - this is sooo ugly. not all variable are part of all kernels and these are at least
# some of the ones I've found to be missing in some. This list may need to be augmented over
# time. The alternative it to have conditional tests on all the printing and there is just
# too much of that.
$tcpData{TcpExt}->{PAWSEstab}= $tcpData{TcpExt}->{DelayedACKs}= $tcpData{TcpExt}->{TW}=0;
$tcpData{TcpExt}->{DelayedACKLost}= $tcpData{TcpExt}->{TCPPrequeued}= $tcpData{TcpExt}->{TCPDirectCopyFromPrequeue}=0;
$tcpData{TcpExt}->{TCPHPHits}= $tcpData{TcpExt}->{TCPPureAcks}= $tcpData{TcpExt}->{TCPHPAcks}=0;
$tcpData{TcpExt}->{TCPDSACKOldSent}=$tcpData{TcpExt}->{TCPAbortOnData}=$tcpData{TcpExt}->{TCPAbortOnClose}=0;
$tcpData{TcpExt}->{TCPSackShiftFallback}=0;
$tcpData{IpExt}->{InMcastPkts}= $tcpData{IpExt}->{InBcastPkts}= $tcpData{IpExt}->{InOctets}=0;
$tcpData{IpExt}->{InMcastOctets}=$tcpData{IpExt}->{InBcastOctets}=$tcpData{IpExt}->{OutMcastPkts}=0;
$tcp{IpExt}->{OutOctets}= $tcpData{IpExt}->{OutMcastOctets}=0;
$tcpData{TcpExt}->{TCPLoss}=$tcpData{TcpExt}->{TCPFastRetrans}=0;
$ipErrors=$icmpErrors=$tcpErrors=$udpErrors=$ipExErrors=$tcpExErrors=0;
# this is here strictly for compatibility with older raw files
$NumTcpFields=65;
for ($i=0; $i<$NumTcpFields; $i++)
{ $tcpValue[$i]=$tcpLast[$i]=0; }
# get ready to process first interval noting '$lastSecs' gets initialized
# when the data file is read in playback mode
$lastSecs[0]=$lastSecs[1]=0 if $playback eq '';
$intFirstSeen=0;
initInterval();
# I n i t ' E x t e n d e d ' V a r i a b l e s
# The current thinking is if someone wants to plot extended variables and
# they haven't been collected (remember the rule that when you report for
# plotting, you always produce what's in -s) we better intialize the results
# variables to all zeros.
for ($i=0; $i<$NumCpus; $i++)
{
$userP[$i]=$niceP[$i]=$sysP[$i]=$idleP[$i]=$totlP[$i]=0;
$irqP[$i]=$softP[$i]=$stealP[$i]=$waitP[$i]=0;
}
# these all need to be initialized in case we use /proc/stats since not all variables
# supplied by that
for ($i=0; $i<<$dskIndexNext; $i++)
{
$dskOps[$i]=$dskTicks[$i]=0;
$dskRead[$i]=$dskReadKB[$i]=$dskReadMrg[$i]=0;
$dskWrite[$i]=$dskWriteKB[$i]=$dskWriteMrg[$i]=0;
$dskRqst[$i]=$dskQueLen[$i]=$dskWait[$i]=$dskSvcTime[$i]=$dskUtil[$i]=0;
}
for ($i=0; $i<$netIndexNext; $i++)
{
$netName[$i]="";
$netRxPkt[$i]=$netTxPkt[$i]= $netRxKB[$i]= $netTxKB[$i]= $netRxErr[$i]=
$netRxDrp[$i]=$netRxFifo[$i]=$netRxFra[$i]= $netRxCmp[$i]= $netRxMlt[$i]=
$netTxErr[$i]=$netTxDrp[$i]= $netTxFifo[$i]=$netTxColl[$i]=$netTxCar[$i]=
$netTxCmp[$i]=$netRxErrs[$i]=$netTxErrs[$i]=0;
}
# Don't forget infiniband
for ($i=0; $i<$NumHCAs; $i++)
{
$ibTxKB[$i]=$ibTx[$i]=$ibRxKB[$i]=$ibRx[$i]=$ibErrorsTot[$i]=0;
}
# if we ever want to map scsi devices to their host/channel/etc, this does it
# for partitions
undef @scsi;
$scsiIndex=0;
foreach $device (split(/\s+/, $ScsiInfo))
{
$scsi[$scsiIndex++]=(split(/:/, $device, 2))[1] if $device=~/DA/;
}
# C o n s t a n t H e a d e r S t u f f
# I suppose for performance it would be good to build all headers once,
# but for now at least do a few pieces.
# get mini date/time header string according to $options but also note these
# don't apply to --top mode
$miniDateTime=""; # so we don't get 'undef' down below
$miniDateTime="Time " if $miniTimeFlag;
$miniDateTime="Date Time " if $miniDateFlag && $options=~/d/;
$miniDateTime="Date Time " if $miniDateFlag && $options=~/D/;
$miniDateTime.=" " if $options=~/m/;
$miniFiller=' ' x length($miniDateTime);
# sometimes we want to shift things 1 space to the left.
$miniFiller1=substr($miniFiller, 0, length($miniFiller)-1);
# If we need two lines, we need to align
$len=length($miniDateTime);
$miniBlanks=sprintf("%${len}s", '');
$interval1Counter=0;
# S l a b S t u f f
$slabIndexNext=0;
$slabDataFlag=0;
undef %slabIndex;
# P r o c e s s S t u f f
$procIndexNext=0;
# I n t e r v a l 2 S t u f f
$interval2Counter=0;
# I n t e r v a l 3 S t u f f
$interval3Counter=0;
$ipmiFile->{pre}=[]; # in case no --envrules specified
$ipmiFile->{post}=[];
$ipmiFile->{ignore}=[];
loadEnvRules() if $subsys=~/E/ || $envTestFile ne '';
# Wasn't sure if this should have been buried in 'loadEnvRules()'
# since they're not actualy 'rules'
if ($envRemap ne '')
{
@envRemaps=split(/,/,$envRemap);
for (my $i=0; $i<@envRemaps; $i++)
{
$envRemaps[$i]=~/\/(.*?)\/(.*?)\//;
$ipmiRemap->[$i]->[1]=$1;
$ipmiRemap->[$i]->[2]=$2;
}
}
# A r c h i t e c t u r e S t u f f
$word32=2**32;
$maxword= ($SrcArch=~/ia64|x86_64/) ? 2**64 : $word32;
return(($version, $datestamp, $timestamp, $timesecs, $timezone, $interval, $recSubsys, $recNfsFilt, $recHeader))
if defined($playfile);
}
# I n i t i a l i z e ' L a s t ' V a r i a b l e s
sub initLast
{
# 0=raw 1=rawp
my $rawType=shift;
# just init slab variables because process ones are all dynamic
if (!defined($rawType) || $rawType)
{
for ($i=0; $i<$NumSlabs; $i++)
{
$slabObjActLast[$i]=$slabObjAllLast[$i]=0;
$slabSlabActLast[$i]=$slabSlabAllLast[$i]=0;
}
return if defined($rawType);
}
# Since dynamically defined need to start clean.
undef(%intrptType);
$ctxtLast=$intrptLast=$procLast=0;
$rpcCCallsLast=$rpcSCallsLast=$rpcBadAuthLast=$rpcBadClntLast=0;
$rpcRetransLast=$rpcCredRefLast=0;
$nfsPktsLast=$nfsUdpLast=$nfsTcpLast=$nfsTcpConnLast=0;
$pageinLast=$pageoutLast=$swapinLast=$swapoutLast=0;
$pagefaultLast=$pagemajfaultLast=0;
$opsLast=$readLast=$readKBLast=$writeLast=$writeKBLast=0;
$memFreeLast=$memUsedLast=$memBufLast=$memCachedLast=0;
$memInactLast=$memSlabLast=$memMapLast=$memAnonLast=$memCommitLast=$memLockedLast=0;
$swapFreeLast=$swapUsedLast=0;
for ($i=0; $i<18; $i++)
{
$nfs2CValuesLast[$i]=0;
$nfs2SValuesLast[$i]=0;
}
for ($i=0; $i<22; $i++)
{
$nfs3CValuesLast[$i]=0;
$nfs3SValuesLast[$i]=0;
}
for ($i=0; $i<59; $i++)
{
$nfs4CValuesLast[$i]=0;
$nfs4SValuesLast[$i]=0;
}
for ($i=0; $i<=$NumCpus; $i++)
{
$userLast[$i]=$niceLast[$i]=$sysLast[$i]=$idleLast[$i]=0;
$waitLast[$i]=$irqLast[$i]=$softLast[$i]=$stealLast[$i]=0;
}
for (my $i=0; $i<$CpuNodes; $i++)
{
$numaStat[$i]->{hitsLast}=$numaStat[$i]->{missLast}=$numaStat[$i]->{forLast}=0;
$numaMem[$i]->{freeLast}= $numaMem[$i]->{usedLast}=$numaMem[$i]->{actLast}=0;
$numaMem[$i]->{inactLast}=$numaMem[$i]->{mapLast}= $numaMem[$i]->{anonLast}=0;
$numaMem[$i]->{lockLast}= $numaMem[$i]->{slabLast}=0;
}
# ...and disks
for ($i=0; $i<$dskIndexNext; $i++)
{
$dskOpsLast[$i]=0;
$dskReadLast[$i]=$dskReadKBLast[$i]=$dskReadMrgLast[$i]=$dskReadTicksLast[$i]=0;
$dskWriteLast[$i]=$dskWriteKBLast[$i]=$dskWriteMrgLast[$i]=$dskWriteTicksLast[$i]=0;
$dskInProgLast[$i]=$dskTicksLast[$i]=$dskWeightedLast[$i]=0;
for ($j=0; $j<11; $j++)
{
$dskFieldsLast[$i][$j]=0;
}
}
for ($i=0; $i<$netIndexNext; $i++)
{
$netRxKBLast[$i]=$netRxPktLast[$i]=$netTxKBLast[$i]=$netTxPktLast[$i]=0;
$netRxErrLast[$i]=$netRxDrpLast[$i]=$netRxFifoLast[$i]=$netRxFraLast[$i]=0;
$netRxCmpLast[$i]=$netRxMltLast[$i]=$netTxErrLast[$i]=$netTxDrpLast[$i]=0;
$netTxFifoLast[$i]=$netTxCollLast[$i]=$netTxCarLast[$i]=$netTxCmpLast[$i]=0;
}
# and interconnect
for ($i=0; $i<$NumXRails; $i++)
{
$elanSendFailLast[$i]=$elanNeterrAtomicLast[$i]=$elanNeterrDmaLast[$i]=0;
$elanRxLast[$i]=$elanRxMBLast[$i]=$elanTxLast[$i]=$elanTxMBLast[$i]=0;
$elanPutLast[$i]=$elanPutMBLast[$i]=$elanGetLast[$i]=$elanGetMBLast[$i]=0;
$elanCompLast[$i]=$elanCompMBLast[$i]=0;
}
# IB
for ($i=0; $i<$NumHCAs; $i++)
{
for ($j=0; $j<16; $j++)
{
# There are 2 ports on an hca, numbered 1 and 2
$ibFieldsLast[$i][1][$j]=$ibFieldsLast[$i][2][$j]=0;
}
}
}
# When a subsys is selected for which this is no possibility of collecting
# data, we must disable it in subsys as well as any --export modules which
# explicitly selects that subsys too
sub disableSubsys
{
my $type= shift;
my $why= shift;
my $unique=shift;
# If user specified --all, they shouldn't see these messages
logmsg("W", "-s$type disabled because $why") if !$allFlag;
$subsys=~s/$type//ig if !defined($unique) || !$unique; # disable using /i if unique not set
$subsys=~s/$type//g if defined($unique) && $unique; # otherwise just disablehe one specified
# Not really sure if need to do this but it certainly can't hurt.
$EFlag=0 if $type=~/E/;
$bFlag=$BFlag=0 if $type=~/b/;
$lFlag=$LFlag=0 if $type=~/l/;
$xFlag=$XFlag=0 if $type=~/x/;
# Now make sure any occurances in s= of an export are disabled too.
for (my $i=0; $i<@expOpts; $i++)
{
if ($expOpts[$i]=~/s=.*$type/i)
{
logmsg('W', "found 's=$type' in lexpr so disabled there too") if !$allFlag;
$expOpts[$i]=~s/$type//ig;
}
}
}
# when playing back lustre data, the indexes on the detail stats may be shifted
# relative to collectl logs in which other OSTs existed. In other words in one
# file one may have "ostY ostZ", in a second "ostX ostZ" and in a third "ostY".
# We need to generate index mappings such that ost1 will always map to 0, ost2
# to 1 and so on.
sub remapLustreNames
{
my $hdrNames=shift;
my $allNames=shift;
my $cltType= shift;
my ($i, $j, $uuid, @hdrTemp, @allTemp, @maps);
# the names as contained in the header are always unique, including ':ost' for
# --lustopt O. However, for --lustopts O reporting, we only want the ost part
# and hence the special treatment. Type=1 used to be meaningful before I realized
# stripping off the ':ost' lead to non-unique names and incorrect remapping.
if ($cltType==2)
{
$hdrNames=~s/\S+:(\S+)/$1/g;
$allNames=~s/\S+:(\S+)/$1/g;
}
print "remapLustrenames() -- Type: $cltType HDR: $hdrNames ALL: $allNames\nREMAPPED: "
if $debug & 8;
if ($hdrNames ne '')
{
@hdrTemp=split(/ /, $hdrNames);
@allTemp=split(/ /, $allNames);
for ($i=0; $i=$NumSlabs)
{
$NumSlabs++;
$slabObjActLast[$i]=$slabObjAllLast[$i]=0;
$slabSlabActLast[$i]=$slabSlabAllLast[$i]=0;
logmsg("W", "New slab created after logging started")
}
# since these are NOT counters, the values are actually totals from which we
# can derive changes from individual entries.
if ($SlabVersion eq '1.1')
{
($slabObjActTot[$i], $slabObjAllTot[$i], $slabObjSize[$i],
$slabSlabActTot[$i], $slabSlabAllTot[$i], $slabPagesPerSlab[$i])=(split(/\s+/, $data))[1..6];
$slabObjPerSlab[$i]=($slabSlabAllTot[$i]) ? $slabObjAllTot[$i]/$slabSlabAllTot[$i] : 0;
}
elsif ($SlabVersion=~/^2/)
{
($slabObjActTot[$i], $slabObjAllTot[$i], $slabObjSize[$i],
$slabObjPerSlab[$i], $slabPagesPerSlab[$i],
$slabSlabActTot[$i], $slabSlabAllTot[$i])=(split(/\s+/, $data))[1..5,13,14];
}
# Total Sizes of objects and slabs
$slabObjActTotB[$i]=$slabObjActTot[$i]*$slabObjSize[$i];
$slabObjAllTotB[$i]=$slabObjAllTot[$i]*$slabObjSize[$i];
$slabSlabActTotB[$i]=$slabSlabActTot[$i]*$slabPagesPerSlab[$i]*$PageSize;
$slabSlabAllTotB[$i]=$slabSlabAllTot[$i]*$slabPagesPerSlab[$i]*$PageSize;
$slabObjAct[$i]= $slabObjActTot[$i]- $slabObjActLast[$i];
$slabObjAll[$i]= $slabObjAllTot[$i]- $slabObjAllLast[$i];
$slabSlabAct[$i]=$slabSlabActTot[$i]-$slabSlabActLast[$i];
$slabSlabAll[$i]=$slabSlabAllTot[$i]-$slabSlabAllLast[$i];
$slabObjActLast[$i]= $slabObjActTot[$i];
$slabObjAllLast[$i]= $slabObjAllTot[$i];
$slabSlabActLast[$i]=$slabSlabActTot[$i];
$slabSlabAllLast[$i]=$slabSlabAllTot[$i];
# Changes in total allocation since last one, noting on first pass it's always 0
my $slabTotMemNow=$slabSlabAllTotB[$i];
my $slabTotMemLast=(defined($slabTotalMemLast{$name})) ? $slabTotalMemLast{$name} : $slabTotMemNow;
$slabTotMemChg[$i]=$slabTotMemNow-$slabTotMemLast;
$slabTotMemPct[$i]=($slabTotMemLast!=0) ? 100*$slabTotMemChg[$i]/$slabTotMemLast : 0;
$slabTotalMemLast{$name}=$slabTotMemNow;
# if --slabopt S, only count slabs whose objects or sizes have changed
# since last interval.
# note -- this is only if !S and the slabs themselves change
if ($slabOpts!~/S/ || $slabSlabAct[$i]!=0 || $slabSlabAll[$i]!=0)
{
$slabObjActTotal+= $slabObjActTot[$i];
$slabObjAllTotal+= $slabObjAllTot[$i];
$slabObjActTotalB+= $slabObjActTot[$i]*$slabObjSize[$i];
$slabObjAllTotalB+= $slabObjAllTot[$i]*$slabObjSize[$i];
$slabSlabActTotal+= $slabSlabActTot[$i];
$slabSlabAllTotal+= $slabSlabAllTot[$i];
$slabSlabActTotalB+=$slabSlabActTot[$i]*$slabPagesPerSlab[$i]*$PageSize;
$slabSlabAllTotalB+=$slabSlabAllTot[$i]*$slabPagesPerSlab[$i]*$PageSize;
$slabNumAct++ if $slabSlabAllTot[$i];
$slabNumTot++;
}
}
else
{
# Note as efficient as if..then..elsif..elsif... but a lot more readable
# and more important, no appreciable difference in processing time
my ($slabname, $datatype, $value)=split(/\s+/, $data);
$slabdata{$slabname}->{objsize}=$value if $datatype=~/^object_/; # object_size
$slabdata{$slabname}->{slabsize}=$value if $datatype=~/^slab_/; # slab_size
$slabdata{$slabname}->{order}=$value if $datatype=~/^or/; # order
$slabdata{$slabname}->{objper}=$value if $datatype=~/^objs/; # objs_per_slab
$slabdata{$slabname}->{objects}=$value if $datatype=~/^objects/;
# This is the second of the ('objects','slabs') tuple
if ($datatype=~/^slabs/)
{
my $numSlabs=$slabdata{$slabname}->{slabs}=$value;
$interval2Print=1;
$slabdata{$slabname}->{avail}=$slabdata{$slabname}->{objper}*$numSlabs;
$slabNumTot+= $numSlabs;
$slabObjAvailTot+=$slabdata{$slabname}->{objper}*$numSlabs;
$slabNumObjTot+= $slabdata{$slabname}->{objects};
$slabUsedTot+= $slabdata{$slabname}->{used}=$slabdata{$slabname}->{slabsize}*$slabdata{$slabname}->{objects};
$slabTotalTot+= $slabdata{$slabname}->{total}=$value*($PageSize<<$slabdata{$slabname}->{order});
# Changes in total allocation since last one, noting on first pass it's always 0
my $slabTotMemNow=$slabdata{$slabname}->{total};
my $slabTotMemLast=(defined($slabTotalMemLast{$slabname})) ? $slabTotalMemLast{$slabname} : $slabTotMemNow;
$slabdata{$slabname}->{memchg}=$slabTotMemNow-$slabTotMemLast;
$slabdata{$slabname}->{mempct}=($slabTotMemLast!=0) ? 100*$slabdata{$slabname}->{memchg}/$slabTotMemLast : 0;
$slabTotalMemLast{$slabname}=$slabTotMemNow;
}
}
}
}
elsif ($subsys=~/b/i && $type=~/^buddy/)
{
my @fields=split(/\s+/, $data);
$buddyNode[$budIndex]=$fields[1];
$buddyZone[$budIndex]=$fields[3];
$buddyNode[$budIndex]=~s/,$//;
for (my $i=0; $i<11; $i++)
{
$buddyInfo[$budIndex][$i]=$fields[$i+4];
$buddyInfoTot[$i]+=$fields[$i+4];
}
$budIndex++;
}
# if user requested -sd, we had to force -sc so we can get 'jiffies'
# 2.6 disk stats may also need cpu to get jiffies for micro calculations
elsif ($subsys=~/c|d/i && $type=~/^cpu/)
{
$type=~/^cpu(\d*)/; # can't do above because second "~=" kills $1
$cpuIndex=($1 ne "") ? $1 : $NumCpus; # only happens in pre 1.7.4
$cpuEnabled[$cpuIndex]=1;
$cpusEnabled++ if $cpuIndex != $NumCpus;
($userNow, $niceNow, $sysNow, $idleNow, $waitNow, $irqNow, $softNow, $stealNow)=split(/\s+/, $data);
$stealNow=0 if !defined($stealNow);
if (!defined($idleNow))
{
incomplete("CPU", $lastSecs[$rawPFlag]);
return;
}
# we don't care about saving raw seconds other than in 'last' variable
# Also note that the total number of jiffies may be needed elsewhere (-s p)
# "wait" doesn't happen unti 2.5, but might as well get ready now.
$user= fix($userNow-$userLast[$cpuIndex]);
$nice= fix($niceNow-$niceLast[$cpuIndex]);
$sys= fix($sysNow-$sysLast[$cpuIndex]);
$idle= fix($idleNow-$idleLast[$cpuIndex]);
$wait= fix($waitNow-$waitLast[$cpuIndex]);
$irq= fix($irqNow-$irqLast[$cpuIndex]);
$soft= fix($softNow-$softLast[$cpuIndex]);
$steal=fix($stealNow-$stealLast[$cpuIndex]);
$total=$user+$nice+$sys+$idle+$wait+$irq+$soft+$steal;
$total=100 if $options=~/n/; # when normalizing, this cancels '100*'
$total=1 if !$total; # has seen to be 0 when interval=0;
$userP[$cpuIndex]= 100*$user/$total;
$niceP[$cpuIndex]= 100*$nice/$total;
$sysP[$cpuIndex]= 100*$sys/$total;
$idleP[$cpuIndex]= 100*$idle/$total;
$waitP[$cpuIndex]= 100*$wait/$total;
$irqP[$cpuIndex]= 100*$irq/$total;
$softP[$cpuIndex]= 100*$soft/$total;
$stealP[$cpuIndex]=100*$steal/$total;
$totlP[$cpuIndex]=$userP[$cpuIndex]+$niceP[$cpuIndex]+
$sysP[$cpuIndex]+$irqP[$cpuIndex]+
$softP[$cpuIndex]+$stealP[$cpuIndex];
$userLast[$cpuIndex]= $userNow;
$niceLast[$cpuIndex]= $niceNow;
$sysLast[$cpuIndex]= $sysNow;
$idleLast[$cpuIndex]= $idleNow;
$waitLast[$cpuIndex]= $waitNow;
$irqLast[$cpuIndex]= $irqNow;
$softLast[$cpuIndex]= $softNow;
$stealLast[$cpuIndex]=$stealNow;
}
elsif ($subsys=~/c/ && $type=~/^load/)
{
($loadAvg1, $loadAvg5, $loadAvg15, $loadProcs)=split(/\s+/, $data);
if (!defined($loadProcs))
{
incomplete("LOAD", $lastSecs[$rawPFlag]);
return;
}
($loadRun, $loadQue)=split(/\//, $loadProcs);
$loadRun--; # never count ourself!
}
elsif ($subsys=~/c/ && $type=~/^procs/)
{
# never include ourselves in count of running processes
$data=(split(/\s+/, $data))[0];
$procsRun=$data-1 if $type=~/^procs_r/;
$procsBlock=$data if $type=~/^procs_b/;
}
elsif ($subsys=~/j/i && $type eq 'int')
{
# exclude any interrupt line that doesn't match a specified filter
return if $intFilt ne '' &&
($intFiltKeep ne '' && $data!~/$intFiltKeep/) ||
($intFiltIgnore ne '' && $data=~/$intFiltIgnore/);
# Note that leading space(s) were removed when we split line above
my ($type, @vals)=split(/\s+/, $data, $cpusEnabled+2);
# If the number of enabled CPUs different than the total, we'll have one
# or more missing columns in /proc/interrupts so do a right shift
if ($cpusEnabled!=$NumCpus)
{
# First move the description up to the last position.
$vals[$NumCpus]=$vals[$cpusEnabled];
# Now right shift all the data into the correct CPU slot
my $index=$NumCpus-1;
for (my $i=$cpusEnabled-1; $i>=0; $i--)
{
# if this CPU disabled, just set its count to 0 and move on to next one
$vals[$index--]=0 if !$cpuEnabled[$index];
$vals[$index]=$vals[$i];
$index--;
}
}
# I n i t i a l i z e ' l a s t ' v a l u e s
# Since I'm not sure if new entries can show up dynamically AND because we
# have to find non-numeric entries so we can initialize them, let's just
# always do our initialization dynamically instead of in initRecord().
$type=~s/:$//;
my $typeSort=($type=~/^\d/) ? sprintf("%03d", $type) : $type;
if (!defined($intrptType{$typeSort}))
{
$intrptType{$typeSort}=1;
if ($type!~/ERR|MIS/)
{
# Pull devicename/time BUT note on earlier kernels for non-numeric types
# these fields aren't always filled in
my ($intType, $intDevices)=split(/\s+/, $vals[$NumCpus], 2);
$intType='' if !defined($intType);
$intDevices='' if !defined($intDevices);
chomp $intDevices;
$intName{$typeSort}=sprintf("%-15s %s", $intType, $intDevices);
if ($type!~/^\d/)
{
$intName{$typeSort}="$intType $intDevices";
$intName{$typeSort}=~s/ interrupts$//;
}
}
if ($type=~/^\d/)
{
# We use array for numeric values and a hash for strings as the array
# access is a little faster expecially as the number of entries grows
# We're also reformatting the modifier so the devices line up...
for (my $i=0; $i<$NumCpus; $i++)
{
$intrptLast[$type]->[$i]=0;
}
}
else
{
for (my $i=0; $i<$NumCpus; $i++)
{
$intrptLast{$type}->[$i]=0;
}
}
}
# M a t h h a p p e n s h e r e
for (my $i=0; $i<$NumCpus; $i++)
{
# If a CPU is disabled, just set it's count to zero.
if ($subsys=~/c/i && !$cpuEnabled[$i])
{
$intrpt[$type]->[$i]=0 if $type=~/^\d/;
$intrpt{$type}->[$i]=0 if $type!~/^\d/;
next;
}
if ($type=~/^\d/)
{
$intrpt[$type]->[$i]=$vals[$i]-$intrptLast[$type]->[$i];
$intrptLast[$type]->[$i]=$vals[$i];
$intrptTot[$i]+=$intrpt[$type]->[$i];
}
# Not sure if other types that only hit cpu0
elsif ($i==0 || ($type ne 'ERR' && $type ne 'MIS'))
{
$intrpt{$type}->[$i]=$vals[$i]-$intrptLast{$type}->[$i];
$intrptLast{$type}->[$i]=$vals[$i];
$intrptTot[$i]+=$intrpt{$type}->[$i];
}
}
}
elsif ($subsys=~/l/i && $type=~/OST_(\d+)/)
{
chomp $data;
$index=$1;
($lustreType, $lustreOps, $lustreBytes)=(split(/\s+/, $data))[0,1,6];
$index=$OstMap[$index] if $playback ne ''; # handles remapping is OSTs change position
#print "IDX: $index, $lustreType, $lustreOps, $lustreBytes\n";
$lustreBytes=0 if $lustreOps==0;
if ($lustreType=~/read/)
{
$lustreReadOpsNow= $lustreOps;
$lustreReadKBytesNow= $lustreBytes/$OneKB;
$lustreReadOps[$index]= fix($lustreReadOpsNow-$lustreReadOpsLast[$index]);
$lustreReadKBytes[$index]= fix($lustreReadKBytesNow-$lustreReadKBytesLast[$index]);
$lustreReadOpsLast[$index]= $lustreReadOpsNow;
$lustreReadKBytesLast[$index]=$lustreReadKBytesNow;
$lustreReadOpsTot+= $lustreReadOps[$index];
$lustreReadKBytesTot+= $lustreReadKBytes[$index];
}
else
{
$lustreWriteOpsNow= $lustreOps;
$lustreWriteKBytesNow= $lustreBytes/$OneKB;
$lustreWriteOps[$index]= fix($lustreWriteOpsNow-$lustreWriteOpsLast[$index]);
$lustreWriteKBytes[$index]= fix($lustreWriteKBytesNow-$lustreWriteKBytesLast[$index]);
$lustreWriteOpsLast[$index]= $lustreWriteOpsNow;
$lustreWriteKBytesLast[$index]=$lustreWriteKBytesNow;
$lustreWriteOpsTot+= $lustreWriteOps[$index];
$lustreWriteKBytesTot+= $lustreWriteKBytes[$index];
}
}
elsif ($subsys=~/l/i && $type=~/OST-b_(\d+):(\d+)/)
{
chomp $data;
$index=$1;
$bufNum=$2;
($lustreBufReadNow, $lustreBufWriteNow)=(split(/\s+/, $data))[1,5];
$index=$OstMap[$index] if $playback ne '';
$lustreBufRead[$index][$bufNum]=fix($lustreBufReadNow-$lustreBufReadLast[$index][$bufNum]);
$lustreBufWrite[$index][$bufNum]=fix($lustreBufWriteNow-$lustreBufWriteLast[$index][$bufNum]);
$lustreBufReadTot[$bufNum]+=$lustreBufRead[$index][$bufNum];
$lustreBufWriteTot[$bufNum]+=$lustreBufWrite[$index][$bufNum];
$lustreBufReadLast[$index][$bufNum]= $lustreBufReadNow;
$lustreBufWriteLast[$index][$bufNum]=$lustreBufWriteNow;
}
elsif ($subsys=~/l/ && $type=~/MDS/)
{
chomp $data;
($name, $value)=(split(/\s+/, $data))[0,1];
# if we ever do mds detail, this goes here!
#$index=$MdsMap[$index] if $playback ne '';
if ($name=~/^mds_getattr$/)
{
$lustreMdsGetattr=fix($value-$lustreMdsGetattrLast);
$lustreMdsGetattrLast=$value;
}
elsif ($name=~/^mds_getattr_lock/)
{
$lustreMdsGetattrLock=fix($value-$lustreMdsGetattrLockLast);
$lustreMdsGetattrLockLast=$value;
}
elsif ($name=~/^mds_statfs/)
{
$lustreMdsStatfs=fix($value-$lustreMdsStatfsLast);
$lustreMdsStatfsLast=$value;
}
elsif ($name=~/^mds_getxattr/)
{
$lustreMdsGetxattr=fix($value-$lustreMdsGetxattrLast);
$lustreMdsGetxattrLast=$value;
}
elsif ($name=~/^mds_setxattr/)
{
$lustreMdsSetxattr=fix($value-$lustreMdsSetxattrLast);
$lustreMdsSetxattrLast=$value;
}
elsif ($name=~/^mds_sync/)
{
$lustreMdsSync=fix($value-$lustreMdsSyncLast);
$lustreMdsSyncLast=$value;
}
elsif ($name=~/^mds_connect/)
{
$lustreMdsConnect=fix($value-$lustreMdsConnectLast);
$lustreMdsConnectLast=$value;
}
elsif ($name=~/^mds_disconnect/)
{
$lustreMdsDisconnect=fix($value-$lustreMdsDisconnectLast);
$lustreMdsDisconnectLast=$value;
}
elsif ($name=~/^mds_reint$/)
{
$lustreMdsReint=fix($value-$lustreMdsReintLast);
$lustreMdsReintLast=$value;
}
# These 5 were added in 1.6.5.1 and are mutually exclusive with mds_reint
elsif ($name=~/^mds_reint_create/)
{
$lustreMdsReintCreate=fix($value-$lustreMdsReintCreateLast);
$lustreMdsReintCreateLast=$value;
}
elsif ($name=~/^mds_reint_link/)
{
$lustreMdsReintLink=fix($value-$lustreMdsReintLinkLast);
$lustreMdsReintLinkLast=$value;
}
elsif ($name=~/^mds_reint_setattr/)
{
$lustreMdsReintSetattr=fix($value-$lustreMdsReintSetattrLast);
$lustreMdsReintSetattrLast=$value;
}
elsif ($name=~/^mds_reint_rename/)
{
$lustreMdsReintRename=fix($value-$lustreMdsReintRenameLast);
$lustreMdsReintRenameLast=$value;
}
elsif ($name=~/^mds_reint_unlink/)
{
$lustreMdsReintUnlink=fix($value-$lustreMdsReintUnlinkLast);
$lustreMdsReintUnlinkLast=$value;
}
}
elsif ($subsys=~/l/i && $type=~/LLITE:(\d+)/)
{
$fs=$1;
chomp $data;
($name, $ops, $value)=(split(/\s+/, $data))[0,1,6];
$fs=$CltFSMap[$fs] if $playback ne '';
if ($name=~/dirty_pages_hits/)
{
$lustreCltDirtyHits[$fs]=fix($ops-$lustreCltDirtyHitsLast[$fs]);
$lustreCltDirtyHitsLast[$fs]=$ops;
$lustreCltDirtyHitsTot+=$lustreCltDirtyHits[$fs];
}
elsif ($name=~/dirty_pages_misses/)
{
$lustreCltDirtyMiss[$fs]=fix($ops-$lustreCltDirtyMissLast[$fs]);
$lustreCltDirtyMissLast[$fs]=$ops;
$lustreCltDirtyMissTot+=$lustreCltDirtyMiss[$fs];
}
elsif ($name=~/read/)
{
# if brand new fs and no I/0, this field isn't defined.
$value=0 if !defined($value);
$lustreCltRead[$fs]=fix($ops-$lustreCltReadLast[$fs]);
$lustreCltReadLast[$fs]=$ops;
$lustreCltReadTot+=$lustreCltRead[$fs];
$lustreCltReadKB[$fs]=fix(($value-$lustreCltReadKBLast[$fs])/$OneKB);
$lustreCltReadKBLast[$fs]=$value;
$lustreCltReadKBTot+=$lustreCltReadKB[$fs];
}
elsif ($name=~/write/)
{
$value=0 if !defined($value); # same as 'read'
$lustreCltWrite[$fs]=fix($ops-$lustreCltWriteLast[$fs]);
$lustreCltWriteLast[$fs]=$ops;
$lustreCltWriteTot+=$lustreCltWrite[$fs];
$lustreCltWriteKB[$fs]=fix(($value-$lustreCltWriteKBLast[$fs])/$OneKB);
$lustreCltWriteKBLast[$fs]=$value;
$lustreCltWriteKBTot+=$lustreCltWriteKB[$fs];
}
elsif ($name=~/open/)
{
$lustreCltOpen[$fs]=fix($ops-$lustreCltOpenLast[$fs]);
$lustreCltOpenLast[$fs]=$ops;
$lustreCltOpenTot+=$lustreCltOpen[$fs];
}
elsif ($name=~/close/)
{
$lustreCltClose[$fs]=fix($ops-$lustreCltCloseLast[$fs]);
$lustreCltCloseLast[$fs]=$ops;
$lustreCltCloseTot+=$lustreCltClose[$fs];
}
elsif ($name=~/seek/)
{
$lustreCltSeek[$fs]=fix($ops-$lustreCltSeekLast[$fs]);
$lustreCltSeekLast[$fs]=$ops;
$lustreCltSeekTot+=$lustreCltSeek[$fs];
}
elsif ($name=~/fsync/)
{
$lustreCltFsync[$fs]=fix($ops-$lustreCltFsyncLast[$fs]);
$lustreCltFsyncLast[$fs]=$ops;
$lustreCltFsyncTot+=$lustreCltFsync[$fs];
}
elsif ($name=~/setattr/)
{
$lustreCltSetattr[$fs]=fix($ops-$lustreCltSetattrLast[$fs]);
$lustreCltSetattrLast[$fs]=$ops;
$lustreCltSetattrTot+=$lustreCltSetattr[$fs];
}
elsif ($name=~/getattr/)
{
$lustreCltGetattr[$fs]=fix($ops-$lustreCltGetattrLast[$fs]);
$lustreCltGetattrLast[$fs]=$ops;
$lustreCltGetattrTot+=$lustreCltGetattr[$fs];
}
}
elsif ($subsys=~/l/i && $type=~/LLITE_RA:(\d+)/)
{
$fs=$1;
chomp $data;
$fs=$CltFSMap[$fs] if $playback ne '';
if ($data=~/^pending.* (\d+)/)
{
# This is NOT a counter but a meter
$ops=$1;
$lustreCltRAPending[$fs]=$ops;
$lustreCltRAPendingTot+=$lustreCltRAPending[$fs];
}
elsif ($data=~/^hits.* (\d+)/)
{
$ops=$1;
$lustreCltRAHits[$fs]=fix($ops-$lustreCltRAHitsLast[$fs]);
$lustreCltRAHitsLast[$fs]=$ops;
$lustreCltRAHitsTot+=$lustreCltRAHits[$fs];
}
elsif ($data=~/^misses.* (\d+)/)
{
$ops=$1;
$lustreCltRAMisses[$fs]=fix($ops-$lustreCltRAMissesLast[$fs]);
$lustreCltRAMissesLast[$fs]=$ops;
$lustreCltRAMissesTot+=$lustreCltRAMisses[$fs];
}
elsif ($data=~/^readpage.* (\d+)/)
{
$ops=$1;
$lustreCltRANotCon[$fs]=fix($ops-$lustreCltRANotConLast[$fs]);
$lustreCltRANotConLast[$fs]=$ops;
$lustreCltRANotConTot+=$lustreCltRANotCon[$fs];
}
elsif ($data=~/^miss inside.* (\d+)/)
{
$ops=$1;
$lustreCltRAMisWin[$fs]=fix($ops-$lustreCltRAMisWinLast[$fs]);
$lustreCltRAMisWinLast[$fs]=$ops;
$lustreCltRAMisWinTot+=$lustreCltRAMisWin[$fs];
}
elsif ($data=~/^failed grab.* (\d+)/)
{
$ops=$1;
$lustreCltRAFalGrab[$fs]=fix($ops-$lustreCltRAFalGrabLast[$fs]);
$lustreCltRAFalGrabLast[$fs]=$ops;
$lustreCltRAFalGrabTot+=$lustreCltRAFalGrab[$fs];
}
elsif ($data=~/^failed lock.* (\d+)/)
{
$ops=$1;
$lustreCltRALckFail[$fs]=fix($ops-$lustreCltRALckFailLast[$fs]);
$lustreCltRALckFailLast[$fs]=$ops;
$lustreCltRALckFailTot+=$lustreCltRALckFail[$fs];
}
elsif ($data=~/^read but.* (\d+)/)
{
$ops=$1;
$lustreCltRAReadDisc[$fs]=fix($ops-$lustreCltRAReadDiscLast[$fs]);
$lustreCltRAReadDiscLast[$fs]=$ops;
$lustreCltRAReadDiscTot+=$lustreCltRAReadDisc[$fs];
}
elsif ($data=~/^zero length.* (\d+)/)
{
$ops=$1;
$lustreCltRAZeroLen[$fs]=fix($ops-$lustreCltRAZeroLenLast[$fs]);
$lustreCltRAZeroLenPLast[$fs]=$ops;
$lustreCltRAZeroLenTot+=$lustreCltRAZeroLen[$fs];
}
elsif ($data=~/^zero size.* (\d+)/)
{
$ops=$1;
$lustreCltRAZeroWin[$fs]=fix($ops-$lustreCltRAZeroWinLast[$fs]);
$lustreCltRAZeroWinLast[$fs]=$ops;
$lustreCltRAZeroWinTot+=$lustreCltRAZeroWin[$fs];
}
elsif ($data=~/^read-ahead.* (\d+)/)
{
$ops=$1;
$lustreCltRA2Eof[$fs]=fix($ops-$lustreCltRA2EofLast[$fs]);
$lustreCltRA2EofLast[$fs]=$ops;
$lustreCltRA2EofTot+=$lustreCltRA2Eof[$fs];
}
elsif ($data=~/^hit max.* (\d+)/)
{
$ops=$1;
$lustreCltRAHitMax[$fs]=fix($ops-$lustreCltRAHitMaxLast[$fs]);
$lustreCltRAHitMaxLast[$fs]=$ops;
$lustreCltRAHitMaxTot+=$lustreCltRAHitMax[$fs];
}
elsif ($data=~/^wrong.* (\d+)/)
{
$ops=$1;
$lustreCltRAWrong[$fs]=fix($ops-$lustreCltRAWrongLast[$fs]);
$lustreCltRAWrong[$fs]=$ops;
$lustreCltRAWrongTot+=$lustreCltRAWrong[$fs];
}
}
elsif ($subsys=~/l/i && $type=~/LLITE_RPC:(\d+):(\d+)/)
{
chomp $data;
$index=$1;
$bufNum=$2;
($lustreCltRpcReadNow, $lustreCltRpcWriteNow)=(split(/\s+/, $data))[1,5];
$index=$CltOstMap[$index] if $playback ne '';
$lustreCltRpcRead[$index][$bufNum]= fix($lustreCltRpcReadNow-$lustreCltRpcReadLast[$index][$bufNum]);
$lustreCltRpcWrite[$index][$bufNum]=fix($lustreCltRpcWriteNow-$lustreCltRpcWriteLast[$index][$bufNum]);
$lustreCltRpcReadTot[$bufNum]+= $lustreCltRpcRead[$index][$bufNum];
$lustreCltRpcWriteTot[$bufNum]+=$lustreCltRpcWrite[$index][$bufNum];
$lustreCltRpcReadLast[$index][$bufNum]= $lustreCltRpcReadNow;
$lustreCltRpcWriteLast[$index][$bufNum]=$lustreCltRpcWriteNow;
}
elsif ($subsys=~/l/i && $type=~/LLDET:(\d+)/)
{
$ost=$1;
chomp $data;
($name, $ops, $value)=(split(/\s+/, $data))[0,1,6];
$ost=$CltOstMap[$ost] if $playback ne '';
if ($name=~/^read_bytes|ost_r/)
{
$lustreCltLunRead[$ost]=fix($ops-$lustreCltLunReadLast[$ost]);
$lustreCltLunReadLast[$ost]=$ops;
if (defined($value)) # not always defined
{
$lustreCltLunReadKB[$ost]=fix(($value-$lustreCltLunReadKBLast[$ost])/$OneKB);
$lustreCltLunReadKBLast[$ost]=$value;
}
}
elsif ($name=~/^write_bytes|ost_w/)
{
$lustreCltLunWrite[$ost]=fix($ops-$lustreCltLunWriteLast[$ost]);
$lustreCltLunWriteLast[$ost]=$ops;
if (defined($value)) # not always defined
{
$lustreCltLunWriteKB[$ost]=(fix($value-$lustreCltLunWriteKBLast[$ost])/$OneKB);
$lustreCltLunWriteKBLast[$ost]=$value;
}
}
}
# disk stats apply to both MDS and OSTs
elsif ($subsys=~/l/i && $type=~/LUS-d_(\d+):(\d+)/)
{
$lusDisk=$1;
$bufNum= $2;
# The units of 'readB/writeB' are number of 512 byte blocks
# in case partial table [rare], make sure totals go in last bucket.
chomp $data;
($size, $reads, $readB, $writes, $writeB)=split(/\s+/, $data);
$bufNum=$LusMaxIndex if $size=~/^total/;
# Numbers for individual disks
$lusDiskReads[$lusDisk][$bufNum]= fix($reads-$lusDiskReadsLast[$lusDisk][$bufNum]);
$lusDiskReadB[$lusDisk][$bufNum]= fix($readB-$lusDiskReadBLast[$lusDisk][$bufNum]);
$lusDiskWrites[$lusDisk][$bufNum]=fix($writes-$lusDiskWritesLast[$lusDisk][$bufNum]);
$lusDiskWriteB[$lusDisk][$bufNum]=fix($writeB-$lusDiskWriteBLast[$lusDisk][$bufNum]);
#print "BEF DISKTOT[$bufNum] R: $lusDiskReadsTot[$bufNum] W: $lusDiskWritesTot[$bufNum]\n";
# Numbers for ALL disks
$lusDiskReadsTot[$bufNum]+= $lusDiskReads[$lusDisk][$bufNum];
$lusDiskReadBTot[$bufNum]+= $lusDiskReadB[$lusDisk][$bufNum];
$lusDiskWritesTot[$bufNum]+=$lusDiskWrites[$lusDisk][$bufNum];
$lusDiskWriteBTot[$bufNum]+=$lusDiskWriteB[$lusDisk][$bufNum];
#print "AFT DISKTOT[$bufNum] R: $lusDiskReadsTot[$bufNum] W: $lusDiskWritesTot[$bufNum]\n";
$lusDiskReadsLast[$lusDisk][$bufNum]= $reads;
$lusDiskReadBLast[$lusDisk][$bufNum]= $readB;
$lusDiskWritesLast[$lusDisk][$bufNum]=$writes;
$lusDiskWriteBLast[$lusDisk][$bufNum]=$writeB;
#print "DISK[$lusDisk][$bufNum] R: $lusDiskReads[$lusDisk][$bufNum] W: $lusDiskWrites[$lusDisk][$bufNum]\n";
}
elsif ($subsys=~/c/ && $type=~/^intr/)
{
$intrptNow=$data;
$intrpt=fix($intrptNow-$intrptLast);
$intrptLast=$intrptNow;
}
elsif ($subsys=~/c/ && $type=~/^ctx/)
{
$ctxtNow=$data;
$ctxt=fix($ctxtNow-$ctxtLast);
$ctxtLast=$ctxtNow;
}
elsif ($subsys=~/c/ && $type=~/^proce/)
{
$procNow=$data;
$proc=fix($procNow-$procLast);
$procLast=$procNow;
}
elsif ($subsys=~/E/ && $type=~/^ipmi/)
{
$interval3Print=1;
my @fields=split(/,/, $data);
# This very first set removes any entries that are to be ignored, even if valid
for (my $i=0; $i{ignore}}); $i++)
{
my $f1=$ipmiFile->{ignore}->[$i]->{f1};
if ($data=~/$f1/)
{
print "Ignore: $data\n" if $envDebug;
return;
}
}
# These are applied BEFORE the pattern match below
print "$data\n" if $envDebug;
my $premap=$fields[0];
$fields[0]=~s/\.|\///g; # get rid of any '.'s or '/'s
for (my $i=0; $i{pre}}); $i++)
{
my $f1=$ipmiFile->{pre}->[$i]->{f1};
my $f2=$ipmiFile->{pre}->[$i]->{f2};
print "/$f1/$f2/\n" if $envDebug;
# No need paying the price of an eval if not symbols to interpret
if ($f2!~/\$/)
{
$fields[0]=~s/$f1/$f2/;
}
else
{
eval "\$fields[0]=~s/$f1/$f2/";
}
print " Pre-Remapped '$premap' to '$fields[0]'\n"
if $premap ne $fields[0] && $envDebug;
}
# matches: Virtual Fan | Fan n | Fans | xxx FANn | Power Meter
# Not really sure why I need the '\s*' but it won't work without it!
if ($fields[0]=~/^(.*)(fan.*?|temp.*?|power meter.*?)\s*(\d*)(.*)$/i)
{
$prefix= defined($1) ? $1 : '';
$name=$2;
$instance=defined($3) ? $3 : '';
$suffix= defined($4) ? $4 : '';
printf " Prefix: %s Name: %s Instance: %s Suffix: %s\n",
$prefix, $name, $instance, $suffix if $envDebug;
$name=~s/Power Meter/Power/;
$type='fan' if $name=~/fan/i;
$type='temp' if $name=~/temp/i;
$type='power' if $name=~/power/i;
$name=~s/\s+$//;
# If a pattern such as 'Fan1A (xxx)', the suffix will actually be set to '1 (xxx)' so
# make 'xxx' the prefix and everything after the '1' will be dropped later anyway
# Power doesn't have a prefix, at least I haven't found any that do yet.
$prefix=$1 if $fields[0]=~/(^fan|^temp)/i && $suffix=~/\((.*)\)/;
# If an instance, append the first 'word' of the suffix as a modifier
if ($instance ne '')
{
$instance.=$suffix;
$instance=~s/\s.*//;
}
# If a pattern like 'Fan xxx' (note the check for NOT starting with a digit),
# there is no prefix or instance so make it start with 'xxx Fan' for which
# we already have logic for checking it for an instance later on.
$prefix=$1 if $prefix eq '' && $instance eq '' && $suffix=~/(^\D+\S+)/;
# If a prefix, typically something like cpu, sys, virtual, etc.,
# prepend the first letter to the name. If it contains any digits
# and we don't yet have an instance, use that as well.
if ($prefix ne '')
{
$prefix=~/(.{1})[a-z]*(\d*)/i;
$name="$1$name";
$instance=$2 if $instance eq '';
}
# Remove all whitespace
$name=~s/\s+//g;
my $postmap=$fields[0];
for (my $i=0; $i{post}}); $i++)
{
my $f1=$ipmiFile->{post}->[$i]->{f1};
my $f2=$ipmiFile->{post}->[$i]->{f2};
print " Post-Remapped '$postmap' to '$name'\n"
if $name=~s/$f1/$f2/ && $envDebug;
}
my $index;
$index=$envFanIndex++ if $type eq 'fan';
$index=$envTempIndex++ if $type eq 'temp';
$index=0 if $type eq 'power';
$fields[1]=-1 if $fields[1] eq '' || $fields[1] eq 'no reading';
# If any last minute name remapping, this is the place for it
for (my $i=0; defined(@$ipmiRemap) && $i<@{$ipmiRemap}; $i++)
{
my $p1=$ipmiRemap->[$i]->[1];
my $p2=$ipmiRemap->[$i]->[2];
$name=~s/$p1/$p2/;
}
$ipmiData->{$type}->[$index]->{name}= $name;
$ipmiData->{$type}->[$index]->{inst}= $instance;
$ipmiData->{$type}->[$index]->{value}= ($fields[1]!~/h$/) ? $fields[1] : $fields[3];
$ipmiData->{$type}->[$index]->{status}=$fields[3];
# we may need to convert temperatures, but be sure it ignore negative values
if ($name=~/Temp/ && $envOpts=~/[CF]/ && ($ipmiData->{$type}->[$index]->{value} != -1))
{
$ipmiData->{$type}->[$index]->{value}= $ipmiData->{$type}->[$index]->{value}*1.8+32 if $envOpts=~/F/ && $fields[2]=~/C$/;
$ipmiData->{$type}->[$index]->{value}= ($ipmiData->{$type}->[$index]->{value}-32)*5/9 if $envOpts=~/C/ && $fields[2]=~/F$/;
}
# finally, if 'T', truncate final value
$ipmiData->{$type}->[$index]->{value}=int($ipmiData->{$type}->[$index]->{value}) if $envOpts=~/T/;
}
}
elsif ($subsys=~/d/i && $type=~/^disk/)
{
($major, $minor, $diskName, @dskFields)=split(/\s+/, $data);
$diskName=~s/cciss\///;
if (!defined($disks{$diskName}))
{
$dskChangeFlag|=1; # new disk found
# if available indexes use one of them otherwise generate a new one.
if (@dskIndexAvail>0)
{ $dskIndex=pop @dskIndexAvail;}
else
{ $dskIndex=$dskIndexNext++; }
$disks{$diskName}=$dskIndex;
print "new disk $diskName [$major,$minor] with index $dskIndex\n" if !$firstPass && $debug & 1;
# add to ordered list of disks if seen for first time
my $newDisk=1;
foreach my $dsk (@dskOrder)
{
$newDisk=0 if $diskName eq $dsk;
}
push @dskOrder, $diskName if $newDisk;
# by initializing the 'last' variable to the current value, we're assured to report 0s for the first
# interval while teeing up the correct last value for the next interval.
for (my $i=0; $i<11; $i++)
{ $dskFieldsLast[$dskIndex][$i]=$dskFields[$i]; }
}
$dskIndex=$disks{$diskName};
$dskSeen[$dskIndex]=$diskName;
$dskSeenCount++; # faster than looping through to count
# Clarification of field definitions:
# Excellent reference: http://cvs.sourceforge.net/viewcvs.py/linux-vax
# /kernel-2.5/Documentation/iostats.txt?rev=1.1.1.2
# ticks - time in jiffies doing I/O (some utils call it 'r/w-use')
# inprog - I/O's in progress (some utils call it 'running')
# ticks - time actually spent doing I/O (some utils call it 'use')
# aveque - average time in queue (some utils call it 'aveq' or even 'ticks')
$dskRead[$dskIndex]= fix($dskFields[0]-$dskFieldsLast[$dskIndex][0]);
$dskReadMrg[$dskIndex]= fix($dskFields[1]-$dskFieldsLast[$dskIndex][1]);
$dskReadKB[$dskIndex]= fix($dskFields[2]-$dskFieldsLast[$dskIndex][2])/2;
$dskReadTicks[$dskIndex]= fix($dskFields[3]-$dskFieldsLast[$dskIndex][3]);
$dskWrite[$dskIndex]= fix($dskFields[4]-$dskFieldsLast[$dskIndex][4]);
$dskWriteMrg[$dskIndex]= fix($dskFields[5]-$dskFieldsLast[$dskIndex][5]);
$dskWriteKB[$dskIndex]= fix($dskFields[6]-$dskFieldsLast[$dskIndex][6])/2;
$dskWriteTicks[$dskIndex]=fix($dskFields[7]-$dskFieldsLast[$dskIndex][7]);
$dskInProg[$dskIndex]= $dskFieldsLast[$dskIndex][8];
$dskTicks[$dskIndex]= fix($dskFields[9]-$dskFieldsLast[$dskIndex][9]);
# according to the author of iostat this field can sometimes be negative
# so handle the same way he does
$dskWeighted[$dskIndex]=($dskFields[10]>=$dskFieldsLast[$dskIndex][10]) ?
fix($dskFields[10]-$dskFieldsLast[$dskIndex][10]) :
fix($dskFieldsLast[$dskIndex][10]-$dskFields[10]);
# If read/write had bogus value, reset ALL current values for this disk to 0, noting that 1st pass
# is initialization and numbers NOT valid so don't generate message
if ($DiskMaxValue>0 && ($dskReadKB[$dskIndex]>$DiskMaxValue || $dskWriteKB[$dskIndex]>$DiskMaxValue))
{
logmsg('E', "One of ReadKB/WriteKB of '$dskRead[$dskIndex]/$dskWriteKB[$dskIndex]' > '$DiskMaxValue' for '$diskName'")
if !$firstPass;
logmsg('W', "Resetting all current performance values for this disk to 0");
$dskOps[$dskIndex]=$dskRead[$dskIndex]=$dskReadKB[$dskIndex]=$dskWrite[$dskIndex]=$dskWriteKB[$dskIndex]=0;
$dskReadMrg[$dskIndex]=$dskWriteMrg[$dskIndex]=$dskWriteTicks[$dskIndex]=0;
$dskInProg[$dskIndex]=$dskTicks[$dskIndex]=$dskWeighted[$dskIndex]=0;
}
# Apply filters to summary totals, explicitly ignoring those we don't want
if ($diskName!~/^dm-|^psv/ && ($dskFilt eq '' || $diskName!~/$dskFiltIgnore/))
{
# if some explicitly named to keep, keep only those
if ($dskFiltKeep eq '' || $diskName=~/$dskFiltKeep/)
{
$dskReadTot+= $dskRead[$dskIndex];
$dskReadMrgTot+= $dskReadMrg[$dskIndex];
$dskReadKBTot+= $dskReadKB[$dskIndex];
$dskReadTicksTot+= $dskReadTicks[$dskIndex];
$dskWriteTot+= $dskWrite[$dskIndex];
$dskWriteMrgTot+= $dskWriteMrg[$dskIndex];
$dskWriteKBTot+= $dskWriteKB[$dskIndex];
$dskWriteTicksTot+=$dskWriteTicks[$dskIndex];
}
}
# needed for compatibility with 2.4 in -P output
$dskOpsTot=$dskReadTot+$dskWriteTot;
# get interval time in jiffies and if no hires, just fudge it noting
# it's pretty rare that hiRes::Time isn't available
$microInterval=($hiResFlag) ? ($fullTime-$lastSecs[$rawPFlag])*100 : $interval*100;
$numIOs=$dskRead[$dskIndex]+$dskWrite[$dskIndex];
$dskRqst[$dskIndex]= $numIOs ? ($dskReadKB[$dskIndex]+$dskWriteKB[$dskIndex])/$numIOs : 0;
$dskQueLen[$dskIndex]= $dskTicks[$dskIndex] ? $dskWeighted[$dskIndex]/$dskTicks[$dskIndex] : 0;
$dskWait[$dskIndex]= $numIOs ? ($dskReadTicks[$dskIndex]+$dskWriteTicks[$dskIndex])/$numIOs : 0;
$dskSvcTime[$dskIndex]=$numIOs ? $dskTicks[$dskIndex]/$numIOs : 0;
$dskUtil[$dskIndex]= $dskTicks[$dskIndex]*10/$microInterval;
# note fieldsLast[8] ignored
for ($i=0; $i<11; $i++)
{
$dskFieldsLast[$dskIndex][$i]=$dskFields[$i];
}
$dskIndex++;
}
elsif ($subsys=~/i/ && $type=~/^fs-/)
{
if ($type=~/^fs-ds/)
{
($dentryNum, $dentryUnused)=(split(/\s+/, $data))[0,1];
}
elsif ($type=~/^fs-fnr/)
{
($filesAlloc, $filesMax)=(split(/\s+/, $data))[0,2];
}
elsif ($type=~/^fs-is/)
{
($inodeMax, $inodeUsed)=split(/\s+/, $data);
$inodeUsed=$inodeMax-$inodeUsed;
}
}
# Only if collecting nfs server data
elsif ($subsys=~/f/i && $type=~/^nfs(.?)-net/)
{
# for earlier versions we've already know this is server data
# but for newer ones we collectl both
my $nfsType=($1 ne '') ? $1 : 's';
($nfsPktsNow, $nfsUdpNow, $nfsTcpNow, $nfsTcpConnNow)=split(/\s+/, $data);
if (!defined($nfsTcpConnNow))
{
incomplete("NFS-NET", $lastSecs[$rawPFlag]);
return;
}
# only look at the data for servers
if ($nfsType eq 's')
{
$nfsPkts= fix($nfsPktsNow-$nfsPktsLast);
$nfsUdp= fix($nfsUdpNow-$nfsUdpLast);
$nfsTcp= fix($nfsTcpNow-$nfsTcpLast);
$nfsTcpConn=fix($nfsTcpConnNow-$nfsTcpConnLast);
$nfsPktsLast= $nfsPktsNow;
$nfsUdpLast= $nfsUdpNow;
$nfsTcpLast= $nfsTcpNow;
$nfsTcpConnLast=$nfsTcpConnNow;
$nfsUdpTot+=$nfsUdp;
$nfsTcpTot+=$nfsTcp;
$nfsTcpConnsTot+=$nfsTcpConn;
}
}
# nfs rpc for server doesn't use fields 2/5
elsif ($subsys=~/f/i && $type=~/^nfs(.?)-rpc/)
{
# For earlier versions, we've already set the client/server flag
my $nfsType=($nfsCFlag) ? 'c' : 's';
$nfsType=$1 if $1 ne '';
my (@rpcFields)=split(/\s+/, $data);
if (($nfsType eq 'c' && !defined($rpcFields[2])) || ($nfsType eq 's' && !defined($rpcFields[4])))
{
incomplete("RPC", $lastSecs[$rawPFlag]);
return;
}
# NOTE - 'calls' common to both clients and servers
if ($nfsType eq 's' && $nfsSFlag)
{
$rpcCallsNow= $rpcFields[0];
$rpcBadAuthNow=$rpcFields[2];
$rpcBadClntNow=$rpcFields[3];
$rpcSCalls= fix($rpcCallsNow-$rpcSCallsLast);
$rpcBadAuth= fix($rpcBadAuthNow-$rpcBadAuthLast);
$rpcBadClnt= fix($rpcBadClntNow-$rpcBadClntLast);
$rpcSCallsLast= $rpcCallsNow;
$rpcBadAuthLast=$rpcBadAuthNow;
$rpcBadClntLast=$rpcBadClntNow;
}
elsif ($nfsCFlag)
{
$rpcCallsNow= $rpcFields[0];
$rpcRetransNow=$rpcFields[1];
$rpcCredRefNow=$rpcFields[2];
$rpcCCalls= fix($rpcCallsNow-$rpcCCallsLast);
$rpcRetrans=fix($rpcRetransNow-$rpcRetransLast);
$rpcCredRef=fix($rpcCredRefNow-$rpcCredRefLast);
$rpcCCallsLast= $rpcCallsNow;
$rpcRetransLast=$rpcRetransNow;
$rpcCredRefLast=$rpcCredRefNow;
}
$rpcBadAuthTot+=$rpcBadAuth;
$rpcBadClntTot+=$rpcBadClnt;
$rpcRetransTot+=$rpcRetrans;
$rpcCredRefTot+=$rpcCredRef;
}
elsif ($subsys=~/f/i && $type=~/^nfs(.?)-proc2/ && $nfs2Flag)
{
# For earlier versions, we've already set the client/server flag
my $nfsType=($nfsCFlag) ? 'c' : 's';
$nfsType=$1 if $1 ne '';
# field 0 is field count, which we know to be 18
@nfsValuesNow=(split(/\s+/, $data))[1..18];
if (scalar(@nfsValuesNow)<18)
{
incomplete("NFS2", $lastSecs[$rawPFlag]);
return;
}
# Until we've seen a non-zero read/write counter for clients/servers
# we won't even look at or report the other fields
if ($nfsValuesNow[6] || $nfsValuesNow[8])
{
$nfs2CSeen=1 if $nfsType eq 'c';
$nfs2SSeen=1 if $nfsType eq 's';
}
if ($nfsType eq 'c' && $nfs2CFlag && $nfs2CSeen)
{
for ($i=0; $i<18; $i++)
{
$nfs2CValue[$i]=fix($nfsValuesNow[$i]-$nfs2CValuesLast[$i]);
$nfs2CValuesLast[$i]=$nfsValuesNow[$i];
}
$nfs2CNull= $nfs2CValue[0]; $nfs2CGetattr= $nfs2CValue[1];
$nfs2CSetattr=$nfs2CValue[2]; $nfs2CRoot= $nfsC2Value[3];
$nfs2CLookup= $nfs2CValue[4]; $nfs2CReadlink=$nfs2CValue[5];
$nfs2CRead= $nfs2CValue[6]; $nfs2CWrcache= $nfs2CValue[7];
$nfs2CWrite= $nfs2CValue[8]; $nfs2CCreate= $nfs2CValue[9];
$nfs2CRemove= $nfs2CValue[10]; $nfs2CRename= $nfs2CValue[11];
$nfs2CLink= $nfs2CValue[12]; $nfs2CSymlink= $nfs2CValue[13];
$nfs2CMkdir= $nfs2CValue[14]; $nfs2CRmdir= $nfs2CValue[15];
$nfs2CReaddir=$nfs2CValue[16]; $nfs2CFsstat= $nfs2CValue[17];
$nfs2CMeta=$nfs2CLookup+$nfs2CSetattr+$nfs2CGetattr+$nfs2CReaddir;
$nfsReadsTot+= $nfs2CRead;
$nfsCReadsTot+= $nfs2CRead;
$nfsWritesTot+= $nfs2CWrite;
$nfsCWritesTot+=$nfs2CWrite;
$nfsMetaTot+= $nfs2CMeta;
$nfsCMetaTot+= $nfs2CMeta;
}
elsif ($nfsType eq 's' && $nfs2SFlag && $nfs2SSeen)
{
for ($i=0; $i<18; $i++)
{
$nfs2SValue[$i]=fix($nfsValuesNow[$i]-$nfs2SValuesLast[$i]);
$nfs2SValuesLast[$i]=$nfsValuesNow[$i];
}
$nfs2SNull= $nfs2SValue[0]; $nfs2SGetattr= $nfs2SValue[1];
$nfs2SSetattr=$nfs2SValue[2]; $nfs2SRoot= $nfs2SValue[3];
$nfs2SLookup= $nfs2SValue[4]; $nfs2SReadlink=$nfs2SValue[5];
$nfs2SRead= $nfs2SValue[6]; $nfs2SWrcache= $nfs2SValue[7];
$nfs2SWrite= $nfs2SValue[8]; $nfs2SCreate= $nfs2SValue[9];
$nfs2SRemove= $nfs2SValue[10]; $nfs2SRename= $nfs2SValue[11];
$nfs2SLink= $nfs2SValue[12]; $nfs2SSymlink= $nfs2SValue[13];
$nfs2SMkdir= $nfs2SValue[14]; $nfs2SRmdir= $nfs2SValue[15];
$nfs2SReaddir=$nfs2SValue[16]; $nfs2SFsstat= $nfs2SValue[17];
$nfs2SMeta=$nfs2SLookup+$nfs2SSetattr+$nfs2SGetattr+$nfs2SReaddir;
$nfsReadsTot+= $nfs2SRead;
$nfsSReadsTot+= $nfs2SRead;
$nfsWritesTot+= $nfs2SWrite;
$nfsSWritesTot+=$nfs2SWrite;
$nfsMetaTot+= $nfs2SMeta;
$nfsSMetaTot+= $nfs2SMeta;
}
}
elsif ($subsys=~/f/i && $type=~/^nfs(.?)-proc3/ && $nfs3Flag)
{
# For earlier versions, we've already set the client/server flag
my $nfsType=($nfsCFlag) ? 'c' : 's';
$nfsType=$1 if $1 ne '';
# field 0 is field count
@nfsValuesNow=(split(/\s+/, $data))[1..22];
if (scalar(@nfsValuesNow)<22)
{
incomplete("NFS3", $lastSecs[$rawPFlag]);
return;
}
if ($nfsValuesNow[6] || $nfsValuesNow[7])
{
$nfs3CSeen=1 if $nfsType eq 'c';
$nfs3SSeen=1 if $nfsType eq 's';
}
if ($nfsType eq 'c' && $nfs3CFlag && $nfs3CSeen)
{
for ($i=0; $i<22; $i++)
{
$nfs3CValue[$i]=fix($nfsValuesNow[$i]-$nfs3CValuesLast[$i]);
$nfs3CValuesLast[$i]=$nfsValuesNow[$i];
}
$nfs3CNull= $nfs3CValue[0]; $nfs3CGetattr= $nfs3CValue[1];
$nfs3CSetattr= $nfs3CValue[2]; $nfs3CLookup= $nfs3CValue[3];
$nfs3CAccess= $nfs3CValue[4]; $nfs3CReadlink= $nfs3CValue[5];
$nfs3CRead= $nfs3CValue[6]; $nfs3CWrite= $nfs3CValue[7];
$nfs3CCreate= $nfs3CValue[8]; $nfs3CMkdir= $nfs3CValue[9];
$nfs3CSymlink= $nfs3CValue[10]; $nfs3CMknod= $nfs3CValue[11];
$nfs3CRemove= $nfs3CValue[12]; $nfs3CRmdir= $nfs3CValue[13];
$nfs3CRename= $nfs3CValue[14]; $nfs3CLink= $nfs3CValue[15];
$nfs3CReaddir= $nfs3CValue[16]; $nfs3CReaddirplus=$nfs3CValue[17];
$nfs3CFsstat= $nfs3CValue[18]; $nfs3CFsinfo= $nfs3CValue[19];
$nfs3CPathconf=$nfs3CValue[20]; $nfs3CCommit= $nfs3CValue[21];
$nfs3CMeta=$nfs3CLookup+$nfs3CAccess+$nfs3CSetattr+$nfs3CGetattr+$nfs3CReaddir+$nfs3CReaddirplus;
$nfsReadsTot+= $nfs3CRead;
$nfsCReadsTot+= $nfs3CRead;
$nfsWritesTot+= $nfs3CWrite;
$nfsCWritesTot+=$nfs3CWrite;
$nfsCommitTot+= $nfs3CCommit;
$nfsCCommitTot+=$nfs3CCommit;
$nfsMetaTot+= $nfs3CMeta;
$nfsCMetaTot+= $nfs3CMeta;
}
elsif ($nfsType eq 's' && $nfs3SFlag && $nfs3SSeen)
{
for ($i=0; $i<22; $i++)
{
$nfs3SValue[$i]=fix($nfsValuesNow[$i]-$nfs3SValuesLast[$i]);
$nfs3SValuesLast[$i]=$nfsValuesNow[$i];
}
$nfs3SNull= $nfs3SValue[0]; $nfs3SGetattr= $nfs3SValue[1];
$nfs3SSetattr= $nfs3SValue[2]; $nfs3SLookup= $nfs3SValue[3];
$nfs3SAccess= $nfs3SValue[4]; $nfs3SReadlink= $nfs3SValue[5];
$nfs3SRead= $nfs3SValue[6]; $nfs3SWrite= $nfs3SValue[7];
$nfs3SCreate= $nfs3SValue[8]; $nfs3SMkdir= $nfs3SValue[9];
$nfs3SSymlink= $nfs3SValue[10]; $nfs3SMknod= $nfs3SValue[11];
$nfs3SRemove= $nfs3SValue[12]; $nfs3SRmdir= $nfs3SValue[13];
$nfs3SRename= $nfs3SValue[14]; $nfs3SLink= $nfs3SValue[15];
$nfs3SReaddir= $nfs3SValue[16]; $nfs3SReaddirplus=$nfs3SValue[17];
$nfs3SFsstat= $nfs3SValue[18]; $nfs3SFsinfo= $nfs3SValue[19];
$nfs3SPathconf=$nfs3SValue[20]; $nfs3SCommit= $nfs3SValue[21];
$nfs3SMeta=$nfs3SLookup+$nfs3SAccess+$nfs3SSetattr+$nfs3SGetattr+$nfs3SReaddir+$nfs3SReaddirplus;
$nfsReadsTot+= $nfs3SRead;
$nfsSReadsTot+= $nfs3SRead;
$nfsWritesTot+= $nfs3SWrite;
$nfsSWritesTot+=$nfs3SWrite;
$nfsCommitTot+= $nfs3SCommit;
$nfsSCommitTot+=$nfs3SCommit;
$nfsMetaTot+= $nfs3SMeta;
$nfsSMetaTot+= $nfs3SMeta;
}
}
# A little trickier because proc4 has client data but proc4ops has server data
elsif ($subsys=~/f/i && $type=~/^nfs(.?)-proc4/ && $nfs4Flag)
{
# For earlier versions, we've already set the client/server flag
my $nfsType=($nfsCFlag) ? 'c' : 's';
$nfsType=$1 if $1 ne '';
# field 0 is field count
($numFields,@nfsValuesNow)=split(/\s+/, $data);
if (scalar(@nfsValuesNow)<$numFields)
{
incomplete("NFS4", $lastSecs[$rawPFlag]);
return;
}
# I can't believe they didn't use the same field numbers for clients/servers
$nfs4CSeen=1 if $nfsType eq 'c' && ($nfsValuesNow[1] || $nfsValuesNow[2]);
$nfs4SSeen=1 if $nfsType eq 's' && ($nfsValuesNow[25] || $nfsValuesNow[38]);
if ($nfsType eq 'c' && $nfs4CFlag && $nfs4CSeen)
{
for ($i=0; $i<$numFields; $i++)
{
$nfs4CValue[$i]=fix($nfsValuesNow[$i]-$nfs4CValuesLast[$i]);
$nfs4CValuesLast[$i]=$nfsValuesNow[$i];
}
# Not Used: Mkdir Mknod Readdirplus Fsstat Rmdir
$nfs4CNull= $nfs4CValue[0]; $nfs4CRead= $nfs4CValue[1];
$nfs4CWrite= $nfs4CValue[2]; $nfs4CCommit= $nfs4CValue[3];
$nfs4CSetattr= $nfs4CValue[9]; $nfs4CFsinfo= $nfs4CValue[10];
$nfs4CAccess= $nfs4CValue[17]; $nfs4CGetattr= $nfs4CValue[18];
$nfs4CLookup= $nfs4CValue[19]; $nfs4CRemove= $nfs4CValue[21];
$nfs4CRename= $nfs4CValue[22]; $nfs4CLink= $nfs4CValue[23];
$nfs4CSymlink= $nfs4CValue[24]; $nfs4CCreate= $nfs4CValue[25];
$nfs4CPathconf=$nfs4CValue[26]; $nfs4CReadlink=$nfs4CValue[28];
$nfs4CReaddir= $nfs4CValue[29];
$nfs4CMeta=$nfs4CLookup+$nfs4CAccess+$nfs4CSetattr+$nfs4CGetattr+$nfs4CReaddir;
$nfsReadsTot+= $nfs4CRead;
$nfsCReadsTot+= $nfs4CRead;
$nfsWritesTot+= $nfs4CWrite;
$nfsCWritesTot+=$nfs4CWrite;
$nfsCommitTot+= $nfs4CCommit;
$nfsCCommitTot+=$nfs4CCommit;
$nfsMetaTot+= $nfs4CMeta;
$nfsCMetaTot+= $nfs4CMeta;
}
elsif ($type=~/^nfs(.?)-proc4ops/ && $nfs4SFlag && $nfs4SSeen)
{
for ($i=0; $i<$numFields; $i++)
{
$nfs4SValue[$i]=fix($nfsValuesNow[$i]-$nfs4SValuesLast[$i]);
$nfs4SValuesLast[$i]=$nfsValuesNow[$i];
}
# Not Used: Null Pathconf Mkdir Mknod Readdirplus Fsinfo Fsstat Symlink Rmdir
$nfs4SAccess= $nfs4SValue[3]; $nfs4SCommit= $nfs4SValue[5];
$nfs4SCreate= $nfs4SValue[6]; $nfs4SGetattr= $nfs4SValue[9];
$nfs4SLink= $nfs4SValue[11]; $nfs4SLookup= $nfs4SValue[15];
$nfs4SRead= $nfs4SValue[25]; $nfs4SReaddir= $nfs4SValue[26];
$nfs4SReadlink=$nfs4SValue[27]; $nfs4SRemove= $nfs4SValue[28];
$nfs4SRename= $nfs4SValue[29]; $nfs4SSetattr= $nfs4SValue[34];
$nfs4SWrite= $nfs4SValue[38];
$nfs4SMeta=$nfs4SLookup+$nfs4SAccess+$nfs4SSetattr+$nfs4SGetattr+$nfs4SReaddir;
$nfsReadsTot+= $nfs4SRead;
$nfsSReadsTot+= $nfs4SRead;
$nfsWritesTot+= $nfs4SWrite;
$nfsSWritesTot+=$nfs4SWrite;
$nfsCommitTot+= $nfs4SCommit;
$nfsSCommitTot+=$nfs4SCommit;
$nfsMetaTot+= $nfs4SMeta;
$nfsSMetaTot+= $nfs4SMeta;
}
}
# M e m o r y S t a t s
elsif ($subsys=~/m/i && $type=~/^pg|^pswp/)
{
if ($type=~/^pgpgin/)
{
$pageinNow=$data;
$pagein=fix($pageinNow-$pageinLast);
$pageinLast=$pageinNow;
}
elsif ($type=~/^pgpgout/)
{
$pageoutNow=$data;
$pageout=fix($pageoutNow-$pageoutLast);
$pageoutLast=$pageoutNow;
}
elsif ($type=~/^pgfault/)
{
$pagefaultNow=$data;
$pagefault=fix($pagefaultNow-$pagefaultLast);
$pagefaultLast=$pagefaultNow;
}
elsif ($type=~/^pgmaj/)
{
$pagemajfaultNow=$data;
$pagemajfault=fix($pagemajfaultNow-$pagemajfaultLast);
$pagemajfaultLast=$pagemajfaultNow;
}
elsif ($type=~/^pswpin/)
{
$swapinNow=$data;
$swapin=fix($swapinNow-$swapinLast);
$swapinLast=$swapinNow;
}
elsif ($type=~/^pswpout/)
{
$swapoutNow=$data;
$swapout=fix($swapoutNow-$swapoutLast);
$swapoutLast=$swapoutNow;
}
if ($memOpts=~/p/)
{
$pageAllocDma=$data if $type=~/^pgalloc_dma$/;
$pageAllocDma32=$data if $type=~/^pgalloc_dma32/;
$pageAllocNormal=$data if $type=~/^pgalloc_normal/;
$pageAllocMove=$data if $type=~/^pgalloc_move/;
$pageRefillDma=$data if $type=~/^pgrefill_dma$/;
$pageRefillDma32=$data if $type=~/^pgrefill_dma32/;
$pageRefillNormal=$data if $type=~/^pgrefill_normal/;
$pageRefillMove=$data if $type=~/^pgrefill_move/;
$pageFree=$data if $type=~/^pgfree/;
$pageActivate=$data if $type=~/^pgactivate/;
}
if ($memOpts=~/s/)
{
$pageStealDma=$data if $type=~/^pgsteal_dma$/;
$pageStealDma32=$data if $type=~/^pgsteal_dma32/;
$pageStealNormal=$data if $type=~/^pgsteal_normal/;
$pageStealMove=$data if $type=~/^pgsteal_move/;
$pageKSwapDma=$data if $type=~/^pgscan_kswapd_dma$/;
$pageKSwapDma32=$data if $type=~/^pgscan_kswapd_dma32/;
$pageKSwapNormal=$data if $type=~/^pgscan_kswapd_normal/;
$pageKSwapMove=$data if $type=~/^pgscan_kswapd_move/;
$pageDirectDma=$data if $type=~/^pgscan_direct_dma$/;
$pageDirectDma32=$data if $type=~/^pgscan_direct_dma32/;
$pageDirectNormal=$data if $type=~/^pgscan_direct_normal/;
$pageDirectMove=$data if $type=~/^pgscan_direct_move/;
}
}
elsif ($subsys=~/m/i && $type=~/^Mem/)
{
$data=(split(/\s+/, $data))[0];
$memTot= $data if $type=~/^MemTotal/;
if ($type=~/^MemFree/)
{
$memFree=$data;
$memFreeC=$memFree-$memFreeLast;
$memFreeLast=$memFree;
}
}
elsif ($subsys=~/m/i && $type=~/^Buffers|^Cached|^Dirty|^Active|^Inactive|^AnonPages|^Mapped|^Slab:|^Committed_AS:|^Huge|^SUnreclaim|^Mloc/)
{
$data=(split(/\s+/, $data))[0];
$memBuf=$data if $type=~/^Buf/;
$memCached=$data if $type=~/^Cac/;
$memDirty=$data if $type=~/^Dir/;
$memAct=$data if $type=~/^Act/;
$memInact=$data if $type=~/^Ina/;
$memSlab=$data if $type=~/^Sla/;
$memAnon=$data if $type=~/^Anon/;
$memMap=$data if $type=~/^Map/;
$memLocked=$data if $type=~/^Mlocked/;
$memCommit=$data if $type=~/^Com/;
$memHugeTot=$data if $type=~/^HugePages_T/;
$memHugeFree=$data if $type=~/^HugePages_F/;
$memHugeRsvd=$data if $type=~/^HugePages_R/;
$memSUnreclaim=$data if $type=~/^SUnreclaim/;
# These are 'changes' since last interval, both positive/negative
# but we only want to do when last one in list seen.
if ($type=~/^Com/)
{
$memBufC= $memBuf-$memBufLast;
$memCachedC=$memCached-$memCachedLast;
$memInactC= $memInact-$memInactLast;
$memSlabC= $memSlab-$memSlabLast;
$memMapC= $memMap-$memMapLast;
$memAnonC= $memAnon-$memAnonLast;
$memCommitC=$memCommit-$memCommitLast;
$memLockedC=$memLocked-$memLockedLast;
$memBufLast= $memBuf;
$memCachedLast=$memCached;
$memInactLast= $memInact;
$memSlabLast= $memSlab;
$memMapLast= $memMap;
$memAnonLast= $memAnon;
$memCommitLast=$memCommit;
$memLockedLast=$memLocked;
}
}
elsif ($subsys=~/m/i && $type=~/^Swap/)
{
$data=(split(/\s+/, $data))[0];
$swapTotal=$data if $type=~/^SwapT/;
if ($type=~/^SwapF/)
{
$swapFree=$data;
$swapFreeC=$swapFree-$swapFreeLast;
$swapFreeLast=$swapFree;
}
}
elsif ($subsys=~/m/i && $type=~/^numa(\S)/)
{
my $statsType=$1;
$data=~/Node (\d+) (\S+?):*\s+(\d+)/;
my $node=$1;
my $name=$2;
my $value=$3;
if ($statsType eq 'i')
{
if ($name=~/^MemFree/)
{ $numaMem[$node]->{free}=$value; }
elsif ($name=~/^MemUsed/)
{ $numaMem[$node]->{used}=$value; }
elsif ($name=~/^Active$/) # equal to Active(anon) + Active(file)
{ $numaMem[$node]->{act}=$value; }
elsif ($name=~/^Inactive$/) # equal to Inactive(anon) + Inactive(file)
{ $numaMem[$node]->{inact}=$value; }
elsif ($name=~/^Mapped/)
{ $numaMem[$node]->{map}=$value; }
elsif ($name=~/^Anon/)
{ $numaMem[$node]->{anon}=$value; }
elsif ($name=~/^Mlock/)
{ $numaMem[$node]->{lock}=$value; }
# currently the last entry read...
elsif ($name=~/^Slab/)
{
$numaMem[$node]->{slab}=$value;
# these are changed since all last seen
if ($memOpts=~/R/)
{
$numaMem[$node]->{freeC}= $numaMem[$node]->{free}- $numaMem[$node]->{freeLast};
$numaMem[$node]->{usedC}= $numaMem[$node]->{used}- $numaMem[$node]->{usedLast};
$numaMem[$node]->{actC}= $numaMem[$node]->{act}- $numaMem[$node]->{actLast};
$numaMem[$node]->{inactC}=$numaMem[$node]->{inact}-$numaMem[$node]->{inactLast};
$numaMem[$node]->{mapC}= $numaMem[$node]->{map}- $numaMem[$node]->{mapLast};
$numaMem[$node]->{anonC}= $numaMem[$node]->{anon}- $numaMem[$node]->{anonLast};
$numaMem[$node]->{lockC}= $numaMem[$node]->{lock}- $numaMem[$node]->{lockLast};
$numaMem[$node]->{slabC}= $numaMem[$node]->{slab}- $numaMem[$node]->{slabLast};
$numaMem[$node]->{freeLast}= $numaMem[$node]->{free};
$numaMem[$node]->{usedLast}= $numaMem[$node]->{used};
$numaMem[$node]->{actLast}= $numaMem[$node]->{act};
$numaMem[$node]->{inactLast}=$numaMem[$node]->{inact};
$numaMem[$node]->{mapLast}= $numaMem[$node]->{map};
$numaMem[$node]->{anonLast}= $numaMem[$node]->{anon};
$numaMem[$node]->{lockLast}= $numaMem[$node]->{lock};
$numaMem[$node]->{slabLast}= $numaMem[$node]->{slab};
}
}
}
else
{
if ($name=~/^numa_hit/)
{ $numaStat[$node]->{hitsNow}=$value; }
elsif ($name=~/^numa_miss/)
{ $numaStat[$node]->{missNow}=$value; }
# currently last entry processed
elsif ($name=~/^numa_foreign/)
{
$numaStat[$node]->{forNow}=$value;
$numaStat[$node]->{hits}=$numaStat[$node]->{hitsNow}-$numaStat[$node]->{hitsLast};
$numaStat[$node]->{miss}=$numaStat[$node]->{missNow}-$numaStat[$node]->{missLast};
$numaStat[$node]->{for}=$numaStat[$node]->{forNow}-$numaStat[$node]->{forLast};
$numaStat[$node]->{hitsLast}=$numaStat[$node]->{hitsNow};
$numaStat[$node]->{missLast}=$numaStat[$node]->{missNow};
$numaStat[$node]->{forLast}=$numaStat[$node]->{forNow};
}
}
}
# S o c k e t S t a t s
elsif ($subsys=~/s/ && $type=~/^sock/)
{
if ($data=~/^sock/)
{
$data=~/(\d+)$/;
$sockUsed=$1;
}
elsif ($data=~/^TCP/)
{
($sockTcp, $sockOrphan, $sockTw, $sockAlloc, $sockMem)=
(split(/\s+/, $data))[2,4,6,8,10];
}
elsif ($data=~/^UDP/)
{
$data=~/(\d+)$/;
$sockUdp=$1;
}
elsif ($data=~/^RAW/)
{
$data=~/(\d+)$/;
$sockRaw=$1;
}
elsif ($data=~/^FRAG/)
{
$data=~/(\d+).*(\d)$/;
$sockFrag=$1;
$sockFragM=$1;
}
}
# N e t w o r k S t a t s
# a few design notes...
# - %networks is the name of all current networks
# - @netOrder is the discovery order
# - @netIndexAvail is a stack of available, previously used indexes
# - $netIndex is the index assigned the current network being processed
# - $netIndexNext is next available index NOT on @netIndexAvail
# - @netSeen is list of networks seen this interval
# - $netSeenCount is the number of entries in @netSeen
# - $netSeenLast save number seen in last interval
elsif ($subsys=~/n/i && $type=~/^Net/)
{
# insert space after interface if none already there
$data=~s/:(\d)/: $1/;
undef @fields;
@fields=split(/\s+/, $data);
if (@fields<17)
{
incomplete("NET:".$fields[0], $lastSecs[$rawPFlag]);
return;
}
# N e w N e t S e e n
my $netName=$fields[0];
$netName=~s/://;
if (!defined($networks{$netName}))
{
$netChangeFlag|=1; # could be useful to external modules
print "new network found: $netName\n" if !$firstPass && $debug & 1;
# if available indexes use one of them otherwise generate a new one
if (@netIndexAvail>0)
{ $netIndex=pop @netIndexAvail;}
else
{ $netIndex=$netIndexNext++; }
$networks{$netName}=$netIndex;
print "new network $netName with index $netIndex\n" if $debug & 1;
# add to ordered list of networks if seen for first time
my $newNet=1;
foreach my $net (@netOrder)
{
$net=~s/:.*//;
$newNet=0 if $netName eq $net;
}
push @netOrder, $netName if $newNet;
# by initializing the 'last' variable to the current value, we're assured to report 0s for the first
# interval while teeing up the correct last value for the next interval.
$netRxKBLast[$netIndex]= $fields[1];
$netRxPktLast[$netIndex]= $fields[2];
$netRxErrLast[$netIndex]= $fields[3];
$netRxDrpLast[$netIndex]= $fields[4];
$netRxFifoLast[$netIndex]=$fields[5];
$netRxFraLast[$netIndex]= $fields[6];
$netRxCmpLast[$netIndex]= $fields[7];
$netRxMltLast[$netIndex]= $fields[8];
$netTxKBLast[$netIndex]= $fields[9];
$netTxPktLast[$netIndex]= $fields[10];
$netTxErrLast[$netIndex]= $fields[11];
$netTxDrpLast[$netIndex]= $fields[12];
$netTxFifoLast[$netIndex]=$fields[13];
$netTxCollLast[$netIndex]=$fields[14];
$netTxCarLast[$netIndex]= $fields[15];
$netTxCmpLast[$netIndex]= $fields[16];
# won't do anything with speed until we create a new file, but then we'll get a new header
my $line=`find /sys/devices/ 2>&1 | grep net | grep $netName | grep speed`;
$netSpeeds{$netName}='??';
if ($line ne '')
{
$speed=`cat $line 2>&1`;
chomp $speed;
$line=~/.*\/(\S+)\/speed/;
my $netName=$1;
$netSpeeds{$netName}=$speed if $speed=~/Invalid/;
}
# user for bogus speed checks
my $netspeed=($netSpeeds{$netName} ne '??') ? $netSpeeds{$netName} : $DefNetSpeed;
$NetMaxTraffic[$netIndex]=2*$interval*$netspeed*125;
}
$netIndex=$networks{$netName};
$netSeen[$netIndex]=$netName;
$netSeenCount++;
$netNameNow= $fields[0];
$netRxKBNow= $fields[1];
$netRxPktNow= $fields[2];
$netRxErrNow= $fields[3];
$netRxDrpNow= $fields[4];
$netRxFifoNow=$fields[5];
$netRxFraNow= $fields[6];
$netRxCmpNow= $fields[7];
$netRxMltNow= $fields[8];
$netTxKBNow= $fields[9];
$netTxPktNow= $fields[10];
$netTxErrNow= $fields[11];
$netTxDrpNow= $fields[12];
$netTxFifoNow=$fields[13];
$netTxCollNow=$fields[14];
$netTxCarNow= $fields[15];
$netTxCmpNow= $fields[16];
$netRxKB[$netIndex]= fix($netRxKBNow-$netRxKBLast[$netIndex])/1024;
$netTxKB[$netIndex]= fix($netTxKBNow-$netTxKBLast[$netIndex])/1024;
$netRxPkt[$netIndex]=fix($netRxPktNow-$netRxPktLast[$netIndex]);
$netTxPkt[$netIndex]=fix($netTxPktNow-$netTxPktLast[$netIndex]);
# extended/errors
$netRxErr[$netIndex]= fix($netRxErrNow- $netRxErrLast[$netIndex]);
$netRxDrp[$netIndex]= fix($netRxDrpNow- $netRxDrpLast[$netIndex]);
$netRxFifo[$netIndex]=fix($netRxFifoNow-$netRxFifoLast[$netIndex]);
$netRxFra[$netIndex]= fix($netRxFraNow- $netRxFraLast[$netIndex]);
$netRxCmp[$netIndex]= fix($netRxCmpNow- $netRxCmpLast[$netIndex]);
$netRxMlt[$netIndex]= fix($netRxMltNow- $netRxMltLast[$netIndex]);
$netTxErr[$netIndex]= fix($netTxErrNow- $netTxErrLast[$netIndex]);
$netTxDrp[$netIndex]= fix($netTxDrpNow- $netTxDrpLast[$netIndex]);
$netTxFifo[$netIndex]=fix($netTxFifoNow-$netTxFifoLast[$netIndex]);
$netTxColl[$netIndex]=fix($netTxCollNow-$netTxCollLast[$netIndex]);
$netTxCar[$netIndex]= fix($netTxCarNow- $netTxCarLast[$netIndex]);
$netTxCmp[$netIndex]= fix($netTxCmpNow- $netTxCmpLast[$netIndex]);
# It has occasionally been observed that bogus data is returned for some networks.
# If we see anything that looks like twice the typical speed, ignore it but remember
# that during the very first interval this data should be bogus! Also, set ALL data
# points to 0 since we can't trust any of them. Note that the bogus value is now in
# the 'last' variable and so the next valid value will be bogus relative to it, but
# then its value will become 'last' and the following values should be 'happy'.
if ($DefNetSpeed>0 && $intFirstSeen &&
($netRxKB[$netIndex]>$NetMaxTraffic[$netIndex] || $netTxKB[$netIndex]>$NetMaxTraffic[$netIndex]))
{
# we're going through some extra pain to make error messages very explicit. we also can't use
# int() because some bogus values are too big, especially if data collectl on 64 bit machine
# and processed on 32 bit one.
$netTxKB[$netIndex]=~s/\..*//;
$netRxKB[$netIndex]=~s/\..*//;
incomplete("NET:".$netNameNow, $lastSecs[$rawPFlag], 'Bogus');
logmsg('I', "Network speed threshhold: $NetMaxTraffic[$netIndex] Bogus Value(s) - TX: $netTxKB[$netIndex]KB RX: $netRxKB[$netIndex]KB");
my $i=$netIndex;
$netRxKB[$i]=$netTxKB[$i]=$netRxPkt[$i]=$netTxPkt[$i]=0;
$netRxErr[$i]=$netRxDrp[$i]=$netRxFifo[$i]=$netRxFra[$i]=$netRxCmp[$i]=$netRxMlt[$i]=0;
$netTxErr[$i]=$netTxDrp[$i]=$netTxFifo[$i]=$netTxColl[$i]=$netTxCar[$i]=$netTxCmp[$i]=0;
}
# these are derived for simplicity of plotting
$netRxErrs[$netIndex]=$netRxErr[$netIndex]+$netRxDrp[$netIndex]+
$netRxFifo[$netIndex]+$netRxFra[$netIndex];
$netTxErrs[$netIndex]=$netTxErr[$netIndex]+$netTxDrp[$netIndex]+
$netTxFifo[$netIndex]+$netTxColl[$netIndex]+
$netTxCar[$netIndex];
# Ethernet totals only, but no longer using anywhere
if ($netNameNow=~/eth/)
{
$netEthRxKBTot+= $netRxKB[$netIndex];
$netEthRxPktTot+=$netRxPkt[$netIndex];
$netEthTxKBTot+= $netTxKB[$netIndex];
$netEthTxPktTot+=$netTxPkt[$netIndex];
}
# at least for now, we're only worrying about totals on real network
# first, always ignore those in ignore list
if ($netNameNow!~/^lo|^sit|^bond|^vmnet|^vlan|^tap|^dp|^nl/ && ($netFilt eq '' || $netNameNow!~/$netFiltIgnore/))
{
# if filter specified, only include those we want.
# NOTE - we >>>never<<< include aliased networks in the summary calculations
if ($netNameNow!~/\./ && ($netFiltKeep eq '' || $netNameNow=~/$netFiltKeep/))
{
$netRxKBTot+= $netRxKB[$netIndex];
$netRxPktTot+=$netRxPkt[$netIndex];
$netTxKBTot+= $netTxKB[$netIndex];
$netTxPktTot+=$netTxPkt[$netIndex];
$netRxErrTot+= $netRxErr[$netIndex];
$netRxDrpTot+= $netRxDrp[$netIndex];
$netRxFifoTot+=$netRxFifo[$netIndex];
$netRxFraTot+= $netRxFra[$netIndex];
$netRxCmpTot+= $netRxCmp[$netIndex];
$netRxMltTot+= $netRxMlt[$netIndex];
$netTxErrTot+= $netTxErr[$netIndex];
$netTxDrpTot+= $netTxDrp[$netIndex];
$netTxFifoTot+=$netTxFifo[$netIndex];
$netTxCollTot+=$netTxColl[$netIndex];
$netTxCarTot+= $netTxCar[$netIndex];
$netTxCmpTot+= $netTxCmp[$netIndex];
$netRxErrsTot+=$netRxErrs[$netIndex];
$netTxErrsTot+=$netTxErrs[$netIndex];
}
}
$netName[$netIndex]= $netNameNow;
$netRxKBLast[$netIndex]= $netRxKBNow;
$netRxPktLast[$netIndex]=$netRxPktNow;
$netTxKBLast[$netIndex]= $netTxKBNow;
$netTxPktLast[$netIndex]=$netTxPktNow;
$netRxErrLast[$netIndex]=$netRxErrNow;
$netRxDrpLast[$netIndex]=$netRxDrpNow;
$netRxFifoLast[$netIndex]=$netRxFifoNow;
$netRxFraLast[$netIndex]=$netRxFraNow;
$netRxCmpLast[$netIndex]=$netRxCmpNow;
$netRxMltLast[$netIndex]=$netRxMltNow;
$netTxErrLast[$netIndex]=$netTxErrNow;
$netTxDrpLast[$netIndex]=$netTxDrpNow;
$netTxFifoLast[$netIndex]=$netTxFifoNow;
$netTxCollLast[$netIndex]=$netTxCollNow;
$netTxCarLast[$netIndex]=$netTxCarNow;
$netTxCmpLast[$netIndex]=$netTxCmpNow;
}
# N e t w o r k S t a c k S t a t s
# note that even though each line type IS already unique, by including our own type
# we get to skip a bunch of compares when not doing -st
# also note the older versions ignored the IpExt data even though collected so I am too.
elsif ($subsys=~/t/i && $type=~/^tcp-|^Tcp/)
{
# Data comes in pairs, the first line being the headers and the second the data.
# if 'tcp-' present, this is V3.6.4 or more and by removing, we assure old/new data looks the same
# but also, the earlier versions didn't write headers which can change from kernel to kernel!!!
$type=~s/^tcp-//;
$type=~s/:$//;
# NEW TCP STATS
if ($playback eq '' || $recVersion ge '3.6.4')
{
# type always precedes data
if ($data=~/^\d/)
{
my @vals=split(/\s+/, $data);
# init 'last' variables here because we don't know how may there are in the normal init section of code
# and since $intFirstSeen does't get cleared until second pass we also need to see if already defined
for (my $i=0; !$intFirstSeen && $i<@vals; $i++)
{ $tcpData{$type}->{last}->[$i]=0 if !defined($tcpData{$type}->{last}->[$i]); }
for (my $i=0; $i<@vals; $i++)
{
my $name=$tcpData{$type}->{hdr}->[$i];
my $value=$vals[$i]-$tcpData{$type}->{last}->[$i];
#print "Seen: $intFirstSeen Type: $type Name: $name I: $i Val: $vals[$i] Last: $tcpData{$type}->{last}->[$i]\n" if $i==0 && $type=~/Icmp/;
$tcpData{$type}->{$name}=$value;
$tcpData{$type}->{last}->[$i]=$vals[$i];
}
# Error summaries for brief/plot data, nothing nothing for IpExt.
if ($briefFlag || $plotFlag ne '')
{
$ipErrors= $tcpData{Ip}->{InHdrErrors}+$tcpData{Ip}->{InAddrErrors}+
$tcpData{Ip}->{InUnknownProtos}+$tcpData{Ip}->{InDiscards}+
$tcpData{Ip}->{OutDiscards}+ $tcpData{Ip}->{ReasmFails}+
$tcpData{Ip}->{FragFails} if $type eq 'Ip';
$tcpErrors= $tcpData{Tcp}->{AttemptFails}+$tcpData{Tcp}->{InErrs} if $type eq 'Tcp';
$udpErrors= $tcpData{Udp}->{NoPorts}+$tcpData{Udp}->{InErrors} if $type eq 'Udp';
$icmpErrors= $tcpData{Icmp}->{InErrors}+$tcpData{Icmp}->{InDestUnreachs}+
$tcpData{Icmp}->{OutErrors} if $type eq 'Icmp';
$tcpExErrors=$tcpData{TcpExt}->{TCPLoss}+$tcpData{TcpExt}->{TCPFastRetrans} if $type eq 'TcpExt';
}
}
# header: only need to grab on the first interval we see
elsif (!$intFirstSeen)
{
my @headers=split(/\s+/, $data);
for (my $i=0; $i<@headers; $i++)
{ $tcpData{$type}->{hdr}->[$i]=$headers[$i]; }
}
}
# OLD TCP STATS
elsif ($type=~/^TcpExt/) # this is the old way, IP header, but no TCP one
{
chomp $data;
@tcpFields=split(/ /, $data);
for ($i=0; $i<$NumTcpFields; $i++)
{
$tcpValue[$i]=fix($tcpFields[$i]-$tcpLast[$i]);
$tcpLast[$i]=$tcpFields[$i];
#print "$i: $tcpValue[$i] ";
}
# store old version data in new version structures even though the positions
# may be wrong for some kernels.
$tcpData{TcpExt}->{TCPPureAcks}= $tcpValue[27];
$tcpData{TcpExt}->{TCPHPAcks}= $tcpValue[28];
$tcpData{TcpExt}->{TCPLoss}= $tcpValue[40];
$tcpData{TcpExt}->{TCPFastRetrans}=$tcpValue[45];
}
}
# E L A N S t a t s
# we have to test the subsys first becaue $1 gets trashed if first
elsif ($subsys=~/x/i && $type=~/^Elan(\d+)/)
{
$i=$1;
if ($XVersion lt '5.20.0')
{
($name, $value)=(split(/\s+/, $data))[0,1] if $XVersion;
}
else
{
($value, $name)=(split(/\s+/, $data))[0,1] if $XVersion;
}
if ($value=~/^Send/ || $name=~/^Send/)
{
($elanSendFail, $elanNeterrAtomic, $elanNeterrDma)=(split(/\s+/, $data))[1,3,5];
$elanSendFail[$i]= fix($elanSendFail-$elanSendFailLast[$i]);
$elanNeterrAtomic[$i]=fix($elanNeterrAtomic-$elanNeterrAtomicLast[$i]);
$elanNeterrDma[$i]= fix($elanNeterrDma-$elanNeterrDmaLast[$i]);
$elanSendFailTot+= $elanSendFail[$i];
$elanNeterrAtomicTot+=$elanNeterrAtomic[$i];
$elanNeterrDmaTot+= $elanNeterrDma[$i];
$elanSendFailLast[$i]= $elanSendFail;
$elanNeterrAtomicLast[$i]=$elanNeterrAtomic;
$elanNeterrDmaLast[$i]= $elanNeterrDma;
}
elsif ($name=~/^Rx/)
{
$elanRx[$i]= fix($value-$elanRxLast[$i]);
$elanRxLast[$i]=$value;
$elanRxTot= $elanRx[$i];
$elanRxFlag=1;
$elanTxFlag=$elanPutFlag=$elanGetFlag=$elanCompFlag=0;
}
elsif ($name=~/^Tx/)
{
$elanTx[$i]= fix($value-$elanTxLast[$i]);
$elanTxLast[$i]=$value;
$elanTxTot= $elanTx[$i];
$elanTxFlag=1;
$elanRxFlag=$elanPutFlag=$elanGetFlag=$elanCompFlag=0;
}
elsif ($name=~/^Put/)
{
$elanPut[$i]= fix($value-$elanPutLast[$i]);
$elanPutLast[$i]=$value;
$elanPutTot= $elanPut[$i];
$elanPutFlag=1;
$elanTxFlag=$elanRxFlag=$elanGetFlag=$elanCompFlag=0;
}
elsif ($name=~/^Get/)
{
$elanGet[$i]= fix($value-$elanGetLast[$i]);
$elanGetLast[$i]=$value;
$elanGetTot= $elanGet[$i];
$elanGetFlag=1;
$elanTxFlag=$elanRxFlag=$elanPutFlag=$elanCompFlag=0;
}
elsif ($name=~/^Comp/)
{
$elanComp[$i]= fix($value-$elanCompLast[$i]);
$elanCompLast[$i]=$value;
$elanCompTot= $elanComp[$i];
$elanCompFlag=1;
$elanTxFlag=$elanRxFlag=$elanPutFlag=$elanGetFlag=0;
}
elsif ($name=~/^MB/)
{
# NOTE - elan reports data in MB but we want it in KB to be
# consistent with other interconects
if ($elanRxFlag)
{
$elanRxMB= fix($value-$elanRxMBLast[$i], $OneMB);
$elanRxMBLast[$i]=$value;
$elanRxKB[$i]= $elanRxMB*1024;
$elanRxKBTot= $elanRxKB[$i];
}
elsif ($elanTxFlag)
{
$elanTxMB= fix($value-$elanTxMBLast[$i], $OneMB);
$elanTxMBLast[$i]=$value;
$elanTxKB[$i]= $elanTxMB*1024;
$elanTxKBTot= $elanTxKB[$i];
}
elsif ($elanPutFlag)
{
$elanPutMB= fix($value-$elanPutMBLast[$i], $OneMB);
$elanPutMBLast[$i]=$value;
$elanPutKB[$i]= $elanPutMB*1024;
$elanPutKBTot= $elanPutKB[$i];
}
elsif ($elanGetFlag)
{
$elanGetMB= fix($value-$elanGetMBLast[$i], $OneMB);
$elanGetMBLast[$i]=$value;
$elanGetKB[$i]= $elanGetMB*1024;
$elanGetKBTot= $elanGetKB[$i];
}
elsif ($elanCompFlag)
{
$elanCompMB= fix($value-$elanCompMBLast[$i], $OneMB);
$elanCompMBLast[$i]=$value;
$elanCompKB[$i]= $elanCompMB*1024;
$elanCompKBTot= $elanCompKB[$i];
}
else
{
logmsg("W", "### Found elan MB without type flag set");
}
}
}
# I n f i n i b a n d S t a t s
# we have to test the subsys first becaue $1 gets trashed if first
elsif ($subsys=~/x/i && $type=~/^ib(\d+)/)
{
$i=$1;
my ($port, @fieldsNow)=(split(/\s+/, $data))[0,4..19];
# Only 1 of the two ports are actually active at any one time
if ($HCAPorts[$i][$port])
{
# Remember which port is active.
$HCAPortActive=$port;
# Calculate values for each field based on 'last' values.
$ibErrorsTot[$i]=0;
for ($j=0; $j<16; $j++)
{
$fields[$j]=fix($fieldsNow[$j]-$ibFieldsLast[$i][$port][$j]);
$ibFieldsLast[$i][$port][$j]=$fieldsNow[$j];
# the first 12 are accumulated as a single error count and ultimately
# reporting as anbsolute number and NOT a rate so don't use 'last'
$ibErrorsTot[$i]+=$fieldsNow[$j] if $j<12;
}
# Do individual counters, noting that the open fabric one has '-port' appended
# and that their values are alredy absolute and not incrementing counters that
# that need to be adjusted agaist previous versions
if ($type=~/^ib(\d+)-(\d)/)
{
$ibTxKB[$i]=$fieldsNow[12]/256;
$ibTx[$i]= $fieldsNow[14];
$ibRxKB[$i]=$fieldsNow[13]/256;
$ibRx[$i]= $fieldsNow[15];
}
else
{
$ibTxKB[$i]=$fields[12]/256;
$ibTx[$i]= $fields[14];
$ibRxKB[$i]=$fields[13]/256;
$ibRx[$i]= $fields[15];
}
$ibTxKBTot+=$ibTxKB[$i];
$ibTxTot+= $ibTx[$i];
$ibRxKBTot+=$ibRxKB[$i];
$ibRxTot+= $ibRx[$i];
$ibErrorsTotTot+=$ibErrorsTot[$i];
}
}
}
# headers for plot formatted data
sub printPlotHeaders
{
my $i;
##############################
# Core Plot Format Headers
##############################
$headersAll='';
$datetime=(!$utcFlag) ? "#Date${SEP}Time${SEP}" : "#UTC${SEP}";
$headers=($filename ne '') ? "$commonHeader$datetime" : $datetime;
if ($subsys=~/c/)
{
$headers.="[CPU]User%${SEP}[CPU]Nice%${SEP}[CPU]Sys%${SEP}[CPU]Wait%${SEP}";
$headers.="[CPU]Irq%${SEP}[CPU]Soft%${SEP}[CPU]Steal%${SEP}[CPU]Idle%${SEP}[CPU]Totl%${SEP}";
$headers.="[CPU]Intrpt$rate${SEP}[CPU]Ctx$rate${SEP}[CPU]Proc$rate${SEP}";
$headers.="[CPU]ProcQue${SEP}[CPU]ProcRun${SEP}[CPU]L-Avg1${SEP}[CPU]L-Avg5${SEP}[CPU]L-Avg15${SEP}";
$headers.="[CPU]RunTot${SEP}[CPU]BlkTot${SEP}";
}
if ($subsys=~/m/)
{
$headers.="[MEM]Tot${SEP}[MEM]Used${SEP}[MEM]Free${SEP}[MEM]Shared${SEP}[MEM]Buf${SEP}[MEM]Cached${SEP}";
$headers.="[MEM]Slab${SEP}[MEM]Map${SEP}[MEM]Anon${SEP}[MEM]Commit${SEP}[MEM]Locked${SEP}"; # always from V1.7.5 forward
$headers.="[MEM]SwapTot${SEP}[MEM]SwapUsed${SEP}[MEM]SwapFree${SEP}[MEM]SwapIn${SEP}[MEM]SwapOut${SEP}";
$headers.="[MEM]Dirty${SEP}[MEM]Clean${SEP}[MEM]Laundry${SEP}[MEM]Inactive${SEP}";
$headers.="[MEM]PageIn${SEP}[MEM]PageOut${SEP}[MEM]PageFaults${SEP}[MEM]PageMajFaults${SEP}";
$headers.="[MEM]HugeTotal${SEP}[MEM]HugeFree${SEP}[MEM]HugeRsvd${SEP}[MEM]SUnreclaim${SEP}";
}
if ($subsys=~/s/)
{
$headers.="[SOCK]Used${SEP}[SOCK]Tcp${SEP}[SOCK]Orph${SEP}[SOCK]Tw${SEP}[SOCK]Alloc${SEP}";
$headers.="[SOCK]Mem${SEP}[SOCK]Udp${SEP}[SOCK]Raw${SEP}[SOCK]Frag${SEP}[SOCK]FragMem${SEP}";
}
if ($subsys=~/n/)
{
$headers.="[NET]RxPktTot${SEP}[NET]TxPktTot${SEP}[NET]RxKBTot${SEP}[NET]TxKBTot${SEP}";
$headers.="[NET]RxCmpTot${SEP}[NET]RxMltTot${SEP}[NET]TxCmpTot${SEP}";
$headers.="[NET]RxErrsTot${SEP}[NET]TxErrsTot${SEP}";
}
if ($subsys=~/d/)
{
$headers.="[DSK]ReadTot${SEP}[DSK]WriteTot${SEP}[DSK]OpsTot${SEP}";
$headers.="[DSK]ReadKBTot${SEP}[DSK]WriteKBTot${SEP}[DSK]KbTot${SEP}";
$headers.="[DSK]ReadMrgTot${SEP}[DSK]WriteMrgTot${SEP}[DSK]MrgTot${SEP}";
}
if ($subsys=~/i/)
{
$headers.="[INODE]NumDentry${SEP}[INODE]openFiles${SEP}[INODE]MaxFile%${SEP}[INODE]used${SEP}";
}
if ($subsys=~/f/)
{
# Alway write client/server fields
$headers.="[NFS]ReadsS${SEP}[NFS]WritesS${SEP}[NFS]MetaS${SEP}[NFS]CommitS${SEP}";
$headers.="[NFS]Udp${SEP}[NFS]Tcp${SEP}[NFS]TcpConn${SEP}[NFS]BadAuth${SEP}[NFS]BadClient${SEP}";
$headers.="[NFS]ReadsC${SEP}[NFS]WritesC${SEP}[NFS]MetaC${SEP}[NFS]CommitC${SEP}";
$headers.="[NFS]Retrans${SEP}[NFS]AuthRef${SEP}";
}
if ($subsys=~/l/)
{
if ($reportMdsFlag)
{
$headers.="[MDS]Getattr${SEP}[MDS]GetattrLock${SEP}[MDS]Statfs${SEP}[MDS]Sync${SEP}";
$headers.="[MDS]Getxattr${SEP}[MDS]Setxattr${SEP}[MDS]Connect${SEP}[MDS]Disconnect${SEP}";
$headers.="[MDS]Reint${SEP}[MDS]Create${SEP}[MDS]Link${SEP}[MDS]Setattr${SEP}";
$headers.="[MDS]Rename${SEP}[MDS]Unlink${SEP}";
}
if ($reportOstFlag)
{
# We always report basic I/O independent of what user selects with --lustopts
$headers.="[OST]Read${SEP}[OST]ReadKB${SEP}[OST]Write${SEP}[OST]WriteKB${SEP}";
if ($lustOpts=~/B/)
{
foreach my $i (@brwBuckets)
{ $headers.="[OSTB]r${i}P${SEP}"; }
foreach my $i (@brwBuckets)
{ $headers.="[OSTB]w${i}P${SEP}"; }
}
}
if ($lustOpts=~/D/)
{
$headers.="[OSTD]Rds${SEP}[OSTD]Rdk${SEP}[OSTD]Wrts${SEP}[OSTD]Wrtk${SEP}";
foreach my $i (@diskBuckets)
{ $headers.="[OSTD]r${i}K${SEP}"; }
foreach my $i (@diskBuckets)
{ $headers.="[OSTD]w${i}K${SEP}"; }
}
if ($reportCltFlag)
{
# 4 different sizes based on whether which value for --lustopts chosen
# NOTE - order IS critical
$headers.="[CLT]Reads${SEP}[CLT]ReadKB${SEP}[CLT]Writes${SEP}[CLT]WriteKB${SEP}";
$headers.="[CLTM]Open${SEP}[CLTM]Close${SEP}[CLTM]GAttr${SEP}[CLTM]SAttr${SEP}[CLTM]Seek${SEP}[CLTM]FSync${SEP}[CLTM]DrtHit${SEP}[CLTM]DrtMis${SEP}"
if $lustOpts=~/M/;
$headers.="[CLTR]Pend${SEP}[CLTR]Hits${SEP}[CLTR]Misses${SEP}[CLTR]NotCon${SEP}[CLTR]MisWin${SEP}[CLTR]FalGrab${SEP}[CLTR]LckFal${SEP}[CLTR]Discrd${SEP}[CLTR]ZFile${SEP}[CLTR]ZerWin${SEP}[CLTR]RA2Eof${SEP}[CLTR]HitMax${SEP}[CLTR]Wrong${SEP}"
if $lustOpts=~/R/;
if ($lustOpts=~/B/)
{
foreach my $i (@brwBuckets)
{ $headers.="[CLTB]r${i}P${SEP}"; }
foreach my $i (@brwBuckets)
{ $headers.="[CLTB]w${i}P${SEP}"; }
}
}
}
if ($subsys=~/x/)
{
my $int=($NumXRails) ? 'ELAN' : 'IB';
$headers.="[$int]InPkt${SEP}[$int]OutPkt${SEP}[$int]InKB${SEP}[$int]OutKB${SEP}[$int]Err${SEP}";
}
if ($subsys=~/t/)
{
# fixed size easier for plotting, keeping Loss & FTrans for historical reasons...
$headers.="[TCP]IpErr${SEP}[TCP]TcpErr${SEP}[TCP]UdpErr${SEP}[TCP]IcmpErr${SEP}[TCP]Loss${SEP}[TCP]FTrans${SEP}";
}
if ($subsys=~/y/)
{
$headers.="[SLAB]ObjInUse${SEP}[SLAB]ObjInUseB${SEP}[SLAB]ObjAll${SEP}[SLAB]ObjAllB${SEP}";
$headers.="[SLAB]InUse${SEP}[SLAB]InUseB${SEP}[SLAB]All${SEP}[SLAB]AllB${SEP}[SLAB]CacheInUse${SEP}[SLAB]CacheTotal${SEP}";
}
if ($subsys=~/b/)
{
for (my $i=0; $i<11; $i++)
{
$headers.=sprintf("[BUD]%dPage%s$SEP", 2**$i, $i==0 ? '' : 's');
}
}
# custom import headers get appended here if doing summary data.
for (my $i=0; $impSummaryFlag && $i<$impNumMods; $i++)
{
&{$impPrintPlot[$i]}(1, \$headers) if $impOpts[$i]=~/s/;
}
# only if at least one core subsystem selected. if not, make sure
# $headersAll contains the date/time in case writing to the terminal
writeData(0, '', \$headers, $LOG, $ZLOG, 'log', \$headersAll) if $coreFlag || $impSummaryFlag;
$headersAll=$headers if !$coreFlag;
#################################
# Non-Core Plot Format Headers
#################################
# here's the deal with these. if writing to files, each file always gets
# their own headers. However, if writing to the terminal we want one long
# string begining with a single date/time AND we don't bother with the
# common header.
$cpuHeaders=$dskHeaders=$envHeaders=$nfsHeaders=$netHeaders='';
$ostHeaders=$mdsHeaders=$cltHeaders=$tcpHeaders=$elanHeaders='';
# Whenever we print a header to a file, we do both the common header
# and date/time. Remember, if we're printing the terminal, this is
# completely ignored by writeData().
$ch=($filename ne '') ? "$commonHeader$datetime" : $datetime;
if ($subsys=~/C/)
{
for ($i=0; $i<$NumCpus; $i++)
{
$cpuHeaders.="[CPU:$i]User%${SEP}[CPU:$i]Nice%${SEP}[CPU:$i]Sys%${SEP}";
$cpuHeaders.="[CPU:$i]Wait%${SEP}[CPU:$i]Irq%${SEP}[CPU:$i]Soft%${SEP}";
$cpuHeaders.="[CPU:$i]Steal%${SEP}[CPU:$i]Idle%${SEP}[CPU:$i]Totl%${SEP}";
$cpuHeaders.="[CPU:$i]Intrpt${SEP}";
}
writeData(0, $ch, \$cpuHeaders, CPU, $ZCPU, 'cpu', \$headersAll);
}
if ($subsys=~/D/ && $options!~/x/)
{
for (my $i=0; $i<@dskOrder; $i++)
{
$dskName=$dskOrder[$i];
next if ($dskFiltKeep eq '' && $dskName=~/$dskFiltIgnore/) || ($dskFiltKeep ne '' && $dskName!~/$dskFiltKeep/);
$temp= "[DSK]Name${SEP}[DSK]Reads${SEP}[DSK]RMerge${SEP}[DSK]RKBytes${SEP}";
$temp.="[DSK]Writes${SEP}[DSK]WMerge${SEP}[DSK]WKBytes${SEP}[DSK]Request${SEP}";
$temp.="[DSK]QueLen${SEP}[DSK]Wait${SEP}[DSK]SvcTim${SEP}[DSK]Util${SEP}";
$temp=~s/DSK/DSK:$dskName/g;
$temp=~s/cciss\///g;
$dskHeaders.=$temp;
}
writeData(0, $ch, \$dskHeaders, DSK, $ZDSK, 'dsk', \$headersAll);
}
if ($subsys=~/E/)
{
foreach $key (sort keys %$ipmiData)
{
for (my $i=0; $i{$key}}); $i++)
{
my $name=$ipmiData->{$key}->[$i]->{name};
my $inst=($key!~/power/ && $ipmiData->{$key}->[$i]->{inst} ne '-1') ? $ipmiData->{$key}->[$i]->{inst} : '';
$envHeaders.=sprintf("[ENV:$name$inst]Speed$SEP") if $key=~/fan/;
$envHeaders.=sprintf("[ENV:$name$inst]Temp$SEP") if $key=~/temp/;
$envHeaders.=sprintf("[ENV:$name]Watts$SEP") if $key=~/power/;
}
}
writeData(0, $ch, \$envHeaders, ENV, $ZENV, 'env', \$headersAll);
}
if ($subsys=~/F/)
{
if ($nfs2CFlag)
{
my $type='NFS:2cd';
$nfsHeaders.="[$type]Read${SEP}[$type]Write${SEP}[$type]Lookup${SEP}[$type]Getattr${SEP}[$type]Setattr${SEP}[$type]Readdir${SEP}";
$nfsHeaders.="[$type]Create${SEP}[$type]Remove${SEP}[$type]Rename${SEP}[$type]Link${SEP}[$type]ReadLink${SEP}[$type]Null${SEP}";
$nfsHeaders.="[$type]Symlink${SEP}[$type]Mkdir${SEP}[$type]Rmdir${SEP}[$type]Fsstat${SEP}";
}
if ($nfs2SFlag)
{
my $type='NFS:2sd';
$nfsHeaders.="[$type]Read${SEP}[$type]Write${SEP}[$type]Lookup${SEP}[$type]Getattr${SEP}[$type]Setattr${SEP}[$type]Readdir${SEP}";
$nfsHeaders.="[$type]Create${SEP}[$type]Remove${SEP}[$type]Rename${SEP}[$type]Link${SEP}[$type]ReadLink${SEP}[$type]Null${SEP}";
$nfsHeaders.="[$type]Symlink${SEP}[$type]Mkdir${SEP}[$type]Rmdir${SEP}[$type]Fsstat${SEP}";
}
if ($nfs3CFlag)
{
my $type='NFS:3cd';
$nfsHeaders.="[$type]Read${SEP}[$type]Write${SEP}[$type]Commit${SEP}[$type]Lookup${SEP}";
$nfsHeaders.="[$type]Access${SEP}[$type]Getattr${SEP}[$type]Setattr${SEP}[$type]Readdir${SEP}";
$nfsHeaders.="[$type]Create${SEP}[$type]Remove${SEP}[$type]Rename${SEP}[$type]Link${SEP}[$type]ReadLink${SEP}[$type]Null${SEP}";
$nfsHeaders.="[$type]Symlink${SEP}[$type]Mkdir${SEP}[$type]Rmdir${SEP}[$type]Fsstat${SEP}";
$nfsHeaders.="[$type]Fsinfo${SEP}[$type]Pathconf${SEP}[$type]Mknod${SEP}[$type]Readdirplus${SEP}";
}
if ($nfs3SFlag)
{
my $type='NFS:3sd';
$nfsHeaders.="[$type]Read${SEP}[$type]Write${SEP}[$type]Commit${SEP}[$type]Lookup${SEP}";
$nfsHeaders.="[$type]Access${SEP}[$type]Getattr${SEP}[$type]Setattr${SEP}[$type]Readdir${SEP}";
$nfsHeaders.="[$type]Create${SEP}[$type]Remove${SEP}[$type]Rename${SEP}[$type]Link${SEP}[$type]ReadLink${SEP}[$type]Null${SEP}";
$nfsHeaders.="[$type]Symlink${SEP}[$type]Mkdir${SEP}[$type]Rmdir${SEP}[$type]Fsstat${SEP}";
$nfsHeaders.="[$type]Fsinfo${SEP}[$type]Pathconf${SEP}[$type]Mknod${SEP}[$type]Readdirplus${SEP}";
}
if ($nfs4CFlag)
{
my $type='NFS:4cd';
$nfsHeaders.="[$type]Read${SEP}[$type]Write${SEP}[$type]Commit${SEP}[$type]Lookup${SEP}";
$nfsHeaders.="[$type]Access${SEP}[$type]Getattr${SEP}[$type]Setattr${SEP}[$type]Readdir${SEP}";
$nfsHeaders.="[$type]Create${SEP}[$type]Remove${SEP}[$type]Rename${SEP}[$type]Link${SEP}[$type]ReadLink${SEP}[$type]Null${SEP}";
$nfsHeaders.="[$type]Symlink${SEP}[$type]Fsinfo${SEP}[$type]Pathconf${SEP}";
}
if ($nfs4SFlag)
{
my $type='NFS:4sd';
$nfsHeaders.="[$type]Read${SEP}[$type]Write${SEP}[$type]Commit${SEP}[$type]Lookup${SEP}";
$nfsHeaders.="[$type]Access${SEP}[$type]Getattr${SEP}[$type]Setattr${SEP}[$type]Readdir${SEP}";
$nfsHeaders.="[$type]Create${SEP}[$type]Remove${SEP}[$type]Rename${SEP}[$type]Link${SEP}[$type]ReadLink${SEP}";
}
writeData(0, $ch, \$nfsHeaders, NFS, $ZNFS, 'nfs', \$headersAll);
}
if ($subsys=~/M/)
{
$numaHeaders='';
for ($i=0; $i<$CpuNodes; $i++)
{
$numaHeaders.="[NUMA:$i]Used${SEP}[NUMA:$i]Free${SEP}[NUMA:$i]Slab${SEP}[NUMA:$i]Mapped${SEP}";
$numaHeaders.="[NUMA:$i]Anon${SEP}[NUMA:$i]Inactive${SEP}[NUMA:$i]Hits${SEP}";
}
writeData(0, $ch, \$numaHeaders, NUMA, $ZNUMA, 'numa', \$headersAll);
}
if ($subsys=~/N/)
{
for (my $i=0; $i<@netOrder; $i++)
{
# remember, order include net speed
$netName=$netOrder[$i];
$netName=~s/:.*//;
next if ($netFiltKeep eq '' && $netName=~/$netFiltIgnore/) || ($netFiltKeep ne '' && $netName!~/$netFiltKeep/);
$temp= "[NET]Name${SEP}[NET]RxPkt${SEP}[NET]TxPkt${SEP}[NET]RxKB${SEP}[NET]TxKB${SEP}";
$temp.="[NET]RxErr${SEP}[NET]RxDrp${SEP}[NET]RxFifo${SEP}[NET]RxFra${SEP}[NET]RxCmp${SEP}[NET]RxMlt${SEP}";
$temp.="[NET]TxErr${SEP}[NET]TxDrp${SEP}[NET]TxFifo${SEP}[NET]TxColl${SEP}[NET]TxCar${SEP}";
$temp.="[NET]TxCmp${SEP}[NET]RxErrs${SEP}[NET]TxErrs${SEP}";
$temp=~s/NET/NET:$netName/g;
$temp=~s/:]/]/g;
$netHeaders.=$temp;
}
writeData(0, $ch, \$netHeaders, NET, $ZNET, 'net', \$headersAll);
}
if ($subsys=~/L/)
{
if ($reportOstFlag)
{
# We always start with this section
# BRW stats are optional, but if there group them together separately.
for ($i=0; $i<$NumOst; $i++)
{
$inst=$lustreOsts[$i];
$ostHeaders.="[OST:$inst]Ost${SEP}[OST:$inst]Read${SEP}[OST:$inst]ReadKB${SEP}[OST:$inst]Write${SEP}[OST:$inst]WriteKB${SEP}";
}
for ($i=0; $lustOpts=~/B/ && $i<$NumOst; $i++)
{
$inst=$lustreOsts[$i];
foreach my $j (@brwBuckets)
{ $ostHeaders.="[OSTB:$inst]r$j${SEP}"; }
foreach my $j (@brwBuckets)
{ $ostHeaders.="[OSTB:$inst]w$j${SEP}"; }
}
writeData(0, $ch, \$ostHeaders, OST, $ZOST, 'ost', \$headersAll);
}
if ($reportCltFlag)
{
$temp='';
if ($lustOpts=~/O/) # client OST details
{
# we always record I/O in one chunk
for ($i=0; $i<$NumLustreCltOsts; $i++)
{
$inst=$lustreCltOsts[$i];
$temp.="[CLT:$inst]FileSys${SEP}[CLT:$inst]Ost${SEP}[CLT:$inst]Reads${SEP}[CLT:$inst]ReadKB${SEP}[CLT:$inst]Writes${SEP}[CLT:$inst]WriteKB${SEP}";
}
# and if specified, brw stats follow
if ($lustOpts=~/B/)
{
for ($i=0; $i<$NumLustreCltOsts; $i++)
{
$inst=$lustreCltOsts[$i];
foreach my $j (@brwBuckets)
{ $temp.="[CLTB:$inst]r${j}P${SEP}"; }
foreach my $j (@brwBuckets)
{ $temp.="[CLTB:$inst]w${j}P${SEP}"; }
}
}
}
else # just fs details
{
# just like with --lustopts O, these three follow each other in groups
for ($i=0; $i<$NumLustreFS; $i++)
{
$inst=$lustreCltFS[$i];
$temp.="[CLT:$inst]FileSys${SEP}[CLT:$inst]Reads${SEP}[CLT:$inst]ReadKB${SEP}[CLT:$inst]Writes${SEP}[CLT:$inst]WriteKB${SEP}";
}
for ($i=0; $lustOpts=~/M/ && $i<$NumLustreFS; $i++)
{
$inst=$lustreCltFS[$i];
$temp.="[CLTM:$inst]Open${SEP}[CLTM:$inst]Close${SEP}[CLTM:$inst]GAttr${SEP}[CLTM:$inst]SAttr${SEP}";
$temp.="[CLTM:$inst]Seek${SEP}[CLTM:$inst]Fsync${SEP}[CLTM:$inst]DrtHit${SEP}[CLTM:$inst]DrtMis${SEP}";
}
for ($i=0; $lustOpts=~/R/ && $i<$NumLustreFS; $i++)
{
$inst=$lustreCltFS[$i];
$temp.="[CLTR:$inst]Pend${SEP}[CLTR:$inst]Hits${SEP}[CLTR:$inst]Misses${SEP}[CLTR:$inst]NotCon${SEP}[CLTR:$inst]MisWin${SEP}[CLTR:$inst]FalGrab${SEP}[CLTR:$inst]LckFal${SEP}";
$temp.="[CLTR:$inst]Discrd${SEP}[CLTR:$inst]ZFile${SEP}[CLTR:$inst]ZerWin${SEP}[CLTR:$inst]RA2Eof${SEP}[CLTR:$inst]HitMax${SEP}[CLTR:$inst]WrongMax${SEP}";
}
}
$cltHeaders.=$temp;
writeData(0, $ch, \$cltHeaders, CLT, $ZCLT, 'clt', \$headersAll);
}
if ($lustOpts=~/D/)
{
$rdHeader="[OSTD]rds${SEP}[OSTD]rdkb${SEP}";
$wrHeader="[OSTD]wrs${SEP}[OSTD]wrkb${SEP}";
foreach my $i (@diskBuckets)
{ $rdHeader.="[OSTD]r${i}K${SEP}"; }
foreach my $i (@diskBuckets)
{ $wrHeader.="[OSTD]w${i}K${SEP}"; }
for ($i=0; $i<$NumLusDisks; $i++)
{
$temp="[OSTD]Disk${SEP}$rdHeader${SEP}$wrHeader";
$temp=~s/OSTD/OSTD:$LusDiskNames[$i]/g;
$blkHeaders.="$temp${SEP}";
}
writeData(0, $ch, \$blkHeaders, BLK, $ZBLK, 'blk', \$headersAll);
}
}
if ($subsys=~/T/)
{
# This is going to be big!!!
for my $type ('Ip', 'Tcp', 'Udp', 'Icmp', 'IpExt', 'TcpExt')
{
next if $type eq 'Ip' && $tcpFilt!~/i/;
next if $type eq 'Tcp' && $tcpFilt!~/t/;
next if $type eq 'Udp' && $tcpFilt!~/u/;
next if $type eq 'Icmp' && $tcpFilt!~/c/;
next if $type eq 'IpExt' && $tcpFilt!~/I/;
next if $type eq 'TcpExt' && $tcpFilt!~/T/;
foreach my $header (@{$tcpData{$type}->{hdr}})
{ $tcpHeaders.="[TCPD]$header$SEP"; }
}
writeData(0, $ch, \$tcpHeaders, TCP, $ZTCP, 'tcp', \$headersAll);
}
if ($subsys=~/X/ && $NumXRails)
{
for ($i=0; $i<$NumXRails; $i++)
{
$elanHeaders.="[ELAN:$i]Rail${SEP}[ELAN:$i]Rx${SEP}[ELAN:$i]Tx${SEP}[ELAN:$i]RxKB${SEP}[ELAN:$i]TxKB${SEP}[ELAN:$i]Get${SEP}[ELAN:$i]Put${SEP}[ELAN:$i]GetKB${SEP}[ELAN:$i]PutKB${SEP}[ELAN:$i]Comp${SEP}[ELAN:$i]CompKB${SEP}[ELAN:$i]SendFail${SEP}[ELAN:$i]Atomic${SEP}[ELAN:$i]DMA${SEP}";
}
writeData(0, $ch, \$elanHeaders, ELN, $ZELN, 'eln', \$headersAll);
}
if ($subsys=~/X/ && $NumHCAs)
{
for ($i=0; $i<$NumHCAs; $i++)
{
$ibHeaders.="[IB:$i]HCA${SEP}[IB:$i]InPkt${SEP}[IB:$i]OutPkt${SEP}[IB:$i]InKB${SEP}[IB:$i]OutKB${SEP}[IB:$i]Err${SEP}";
}
writeData(0, $ch, \$ibHeaders, IB, $ZIB, 'ib', \$headersAll);
}
$budHeaders='';
if ($subsys=~/B/)
{
for (my $i=0; $i<$NumBud; $i++)
{
my $buddyName="$buddyZone[$i]-$buddyNode[$i]";
$budHeaders.="[BUD:$buddyName]Node${SEP}[BUD:$buddyName]Zone${SEP}";
for (my $j=0; $j<11; $j++)
{
$budHeaders.=sprintf("[BUD:$buddyName]%dPage%s$SEP", 2**$j, $j==0 ? '' : 's');
}
}
writeData(0, $ch, \$budHeaders, BUD, $ZBUD, 'bud', \$headersAll);
}
# only make call(s) if respective modules if detail reporting has been requested
for (my $i=0; $impDetailFlag && $i<$impNumMods; $i++)
{
if ($impOpts[$i]=~/d/)
{
my $impHeaders='';
&{$impPrintPlot[$i]}(2, \$impHeaders);
writeData(0, $ch, \$impHeaders, $impText[$i], $impGz[$i], 'imp-$i', \$headersAll);
}
}
# When going to the terminal OR socket we need a final call with no 'data'
# to write. Also note that there is a final separator that needs to be removed.
# It also turns out if doing --export -P, THAT module is responsible for sending
# data over the socket and the plot data ONLY gets written locally.
# Finally, if there is an error writing to a socket, stop trying to record anything else
# as it's probably a broken socket and '!$doneFlag' has been set and we'll exit cleanly
$headersAll=~s/$SEP$//;
if (!$logToFileFlag || ($sockFlag && $export eq ''))
{
return if writeData(1, '', undef, $LOG, undef, undef, \$headersAll)==0;
}
#################################
# Exception File Headers
#################################
if ($options=~/x/i)
{
if ($subsys=~/D/)
{
$dskHeaders="Num${SEP}";
$dskHeaders.="[DISKX]Name${SEP}[DISKX]Reads${SEP}[DISKX]Merged${SEP}[DISKX]KBytes${SEP}[DISKX]Writes${SEP}[DISKX]Merged${SEP}";
$dskHeaders.="[DISKX]KBytes${SEP}[DISKX]Request${SEP}[DISKX]QueLen${SEP}[DISKX]Wait${SEP}[DISKX]SvcTim${SEP}[DISKX]Util\n";
# Since we never write exception data over a socket the last parameter is undef.
writeData(0, $ch, \$dskHeaders, DSKX, $ZDSKX, 'dskx', undef);
}
}
$headersPrinted=1;
}
sub intervalPrint
{
my $seconds=shift;
# If seconds end in .000, $seconds comes across as integer with no $usecs!
($seconds, $usecs)=split(/\./, $seconds);
$usecs='000' if !defined($usecs); # in case user specifies -om
if ($hiResFlag)
{
$usecs=substr("${usecs}00", 0, 3);
$seconds.=".$usecs";
}
# This is causing confusion because this ALWAYS gets incremented even if no
# output, such as when we only interval2 data
$totalCounter++;
my $tempSubsys=$subsys;
$tempSubsys=~s/Y// if $slabAnalOnlyFlag;
$tempSubsys=~s/Z// if $procAnalOnlyFlag;
printPlot($seconds, $usecs) if $plotFlag && ($tempSubsys ne '' || $import ne '');
printTerm($seconds, $usecs) if !$plotFlag && $expName eq '';
procAnalyze($seconds, $usecs) if $procAnalFlag && $interval2Print;
slabAnalyze($seconds, $usecs) if $slabAnalFlag && $interval2Print;
if ($expName ne '')
{
logdiag('export data') if $utimeMask & 1;
&$expName($expOpts);
exit(0) if $showColFlag;
}
}
# anything that needs to be derived should be done only once and this is the place
sub derived
{
$swapUsed=$swapTotal-$swapFree;
$swapUsedC=$swapUsed-$swapUsedLast;
$swapUsedLast=$swapUsed;
$memUsed=$memTot-$memFree;
$memUsedC=$memUsed-$memUsedLast;
$memUsedLast=$memUsed;
}
###########################
# P l o t F o r m a t
###########################
sub printPlot
{
my $seconds=shift;
my $usecs= shift;
my ($datestamp, $time, $hh, $mm, $ss, $mday, $mon, $year, $i, $j);
# We always print some form of date and time in plot format and in the case of
# --utc, it's a single value. Now that I'm pulling out usecs for utc we
# probably don't have to pass it as the second parameter.
$utcSecs=(split(/\./, $seconds))[0];
($ss, $mm, $hh, $mday, $mon, $year)=localtime($seconds);
$date=($options=~/d/) ?
sprintf("%02d/%02d", $mon+1, $mday) :
sprintf("%d%02d%02d", $year+1900, $mon+1, $mday);
$time= sprintf("%02d:%02d:%02d", $hh, $mm, $ss);
my $datetime=(!$utcFlag) ? "$date$SEP$time": $utcSecs;
$datetime.=".$usecs" if $options=~/m/;
# slab detail and processes have their own print routines because they
# do multiple lines of output and can't be mixed with anything else.
# Furthermore, if we're doing -rawtoo, we DON'T generate these files since
# the data is already being recorded in the raw file and we don't want to do
# both
if (!$rawtooFlag && $subsys=~/[YZ]/ && $interval2Print)
{
printPlotSlab($date, $time) if $subsys=~/Y/ && !$slabAnalOnlyFlag;
printPlotProc($date, $time) if $subsys=~/Z/ && !$procAnalOnlyFlag;
return if $subsys=~/^[YZ]$/; # we're done if ONLY printing slabs or processes
}
# Print headers noting that by default $headerRepeat set to 0 for -P. Also note we have to
# get more elaborate for terminal/file-based plot data. On the terminal when HR is 0, we only
# want one header but when going to files we ALWAYS want a new header each day when
# $headersPrinted gets reset to 0.
$interval1Counter++;
printPlotHeaders() if ($headerRepeat==0 && $filename eq '' && $interval1Counter==1) ||
($headerRepeat==0 && $filename ne '' && !$headersPrinted) ||
($headerRepeat>0 && ($interval1Counter % $headerRepeat)==1);
exit(0) if $showColFlag;
#######################
# C O R E D A T A
#######################
my $netErrors=0;
$plot=$oneline='';
if ($coreFlag || $impSummaryFlag)
{
# CPU Data cols
if ($subsys=~/c/)
{
$i=$NumCpus;
$plot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$userP[$i], $niceP[$i], $sysP[$i], $waitP[$i],
$irqP[$i], $softP[$i], $stealP[$i], $idleP[$i], $totlP[$i]);
$plot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%4.2f$SEP%4.2f$SEP%4.2f$SEP%d$SEP%d",
$intrpt/$intSecs, $ctxt/$intSecs, $proc/$intSecs,
$loadQue, $loadRun, $loadAvg1, $loadAvg5, $loadAvg15, $procsRun, $procsBlock);
}
# MEM
if ($subsys=~/m/)
{
$plot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$memTot, $memUsed, $memFree, $memShared, $memBuf, $memCached);
$plot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS", $memSlab, $memMap, $memAnon, $memCommit, $memLocked); # Always from V1.7.5 forward
$plot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$swapTotal, $swapUsed, $swapFree, $swapin/$intSecs, $swapout/$intSecs,
$memDirty, $clean, $laundry, $memInact,
$pagein/$intSecs, $pageout/$intSecs, $pagefault/$intSecs, $pagemajfault/$intSecs);
$plot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS", $memHugeTot, $memHugeFree, $memHugeRsvd, $memSUnreclaim);
}
# SOCKETS
if ($subsys=~/s/)
{
$plot.="$SEP$sockUsed$SEP$sockTcp$SEP$sockOrphan$SEP$sockTw$SEP$sockAlloc";
$plot.="$SEP$sockMem$SEP$sockUdp$SEP$sockRaw$SEP$sockFrag$SEP$sockFragM";
}
# NETWORKS
if ($subsys=~/n/)
{
# NOTE - rx/tx errs are the totals of all error counters
$plot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$netRxPktTot/$intSecs, $netTxPktTot/$intSecs,
$netRxKBTot/$intSecs, $netTxKBTot/$intSecs,
$netRxCmpTot/$intSecs, $netRxMltTot/$intSecs,
$netTxCmpTot/$intSecs, $netRxErrsTot/$intSecs,
$netTxErrsTot/$intSecs);
$netErrors=$netRxErrsTot+$netTxErrsTot;
}
# DISKS
if ($subsys=~/d/)
{
$plot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$dskReadTot/$intSecs, $dskWriteTot/$intSecs, $dskOpsTot/$intSecs,
$dskReadKBTot/$intSecs, $dskWriteKBTot/$intSecs, ($dskReadKBTot+$dskWriteKBTot)/$intSecs,
$dskReadMrgTot/$intSecs, $dskWriteMrgTot/$intSecs, ($dskReadMrgTot+$dskWriteMrgTot)/$intSecs);
}
# INODES
if ($subsys=~/i/)
{
$plot.=sprintf("$SEP%d$SEP%d$SEP%$FS$SEP%d",
$dentryNum, $filesAlloc, $filesMax ? $filesAlloc*100/$filesMax : 0, $inodeUsed);
}
# NFS
if ($subsys=~/f/)
{
$plot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$nfsSReadsTot/$intSecs, $nfsSWritesTot/$intSecs, $nfsSMetaTot/$intSecs,
$nfsSCommitTot/$intSecs, $nfsUdpTot/$intSecs, $nfsTcpTot/$intSecs,
$nfsTcpConnTot/$intSecs, $rpcBadAuthTot/$intSecs, $rpcBadClntTot/$intSecs);
$plot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$nfsCReadsTot/$intSecs, $nfsCWritesTot/$intSecs, $nfsCMetaTot/$intSecs,
$nfsCCommitTot/$intSecs, $rpcRetransTot/$intSecs, $rpcCredRefTot/$intSecs);
}
# Lustre
if ($subsys=~/l/)
{
# MDS goes first since for detail, the OST is variable and if we ever
# do both we want consistency of order. Also note that by reporting all 6
# reints we assure consisency across lustre versions
if ($reportMdsFlag)
{
$mdsReint=$lustreMdsReintCreate+$lustreMdsReintLink+
$lustreMdsReintSetattr+$lustreMdsReintRename+$lustreMdsReintUnlink
if $cfsVersion lt '1.6.5';
$plot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$lustreMdsGetattr/$intSecs, $lustreMdsGetattrLock/$intSecs,
$lustreMdsStatfs/$intSecs, $lustreMdsSync/$intSecs,
$lustreMdsGetxattr/$intSecs, $lustreMdsSetxattr/$intSecs,
$lustreMdsConnect/$intSecs, $lustreMdsDisconnect/$intSecs,
$lustreMdsReint/$intSecs,
$lustreMdsReintCreate/$intSecs, $lustreMdsReintLink/$intSecs,
$lustreMdsReintSetattr/$intSecs, $lustreMdsReintRename/$intSecs,
$lustreMdsReintUnlink/$intSecs);
}
if ($reportOstFlag)
{
# We always do this...
$plot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$lustreReadOpsTot/$intSecs, $lustreReadKBytesTot/$intSecs,
$lustreWriteOpsTot/$intSecs, $lustreWriteKBytesTot/$intSecs);
if ($lustOpts=~/B/)
{
for ($j=0; $j<$numBrwBuckets; $j++)
{
$plot.=sprintf("$SEP%$FS", $lustreBufReadTot[$j]/$intSecs);
}
for ($j=0; $j<$numBrwBuckets; $j++)
{
$plot.=sprintf("$SEP%$FS", $lustreBufWriteTot[$j]/$intSecs);
}
}
}
# Disk Block Level Stats can apply to both MDS and OST
if ($lustOpts=~/D/)
{
$plot.=sprintf("$SEP%d$SEP%d$SEP%d$SEP%d",
$lusDiskReadsTot[$LusMaxIndex]/$intSecs,
$lusDiskReadBTot[$LusMaxIndex]*0.5/$intSecs,
$lusDiskWritesTot[$LusMaxIndex]/$intSecs,
$lusDiskWriteBTot[$LusMaxIndex]*0.5/$intSecs);
for ($i=0; $i<$LusMaxIndex; $i++)
{ $plot.=sprintf("$SEP%d", $lusDiskReadsTot[$i]/$intSecs); }
for ($i=0; $i<$LusMaxIndex; $i++)
{ $plot.=sprintf("$SEP%d", $lusDiskWritesTot[$i]/$intSecs); }
}
if ($reportCltFlag)
{
# There are actually 3 different formats depending on --lustopts
$plot.=sprintf("$SEP%d$SEP%d$SEP%d$SEP%d",
$lustreCltReadTot/$intSecs, $lustreCltReadKBTot/$intSecs,
$lustreCltWriteTot/$intSecs, $lustreCltWriteKBTot/$intSecs);
$plot.=sprintf("$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d",
$lustreCltOpenTot/$intSecs, $lustreCltCloseTot/$intSecs,
$lustreCltGetattrTot/$intSecs, $lustreCltSetattrTot/$intSecs,
$lustreCltSeekTot/$intSecs, $lustreCltFsyncTot/$intSecs,
$lustreCltDirtyHitsTot/$intSecs, $lustreCltDirtyMissTot/$intSecs)
if $lustOpts=~/M/;
$plot.=sprintf("$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d",
$lustreCltRAPendingTot, $lustreCltRAHitsTot, $lustreCltRAMissesTot,
$lustreCltRANotConTot, $lustreCltRAMisWinTot, $lustreCltRAFalGrabTot,
$lustreCltRALckFailTot, $lustreCltRAReadDiscTot, $lustreCltRAZeroLenTot,
$lustreCltRAZeroWinTot, $lustreCltRA2EofTot, $lustreCltRAHitMaxTot,
$lustreCltRAWrongTot)
if $lustOpts=~/R/;
if ($lustOpts=~/B/) {
for ($i=0; $i<$numBrwBuckets; $i++) {
$plot.=sprintf("$SEP%d", $lustreCltRpcReadTot[$i]/$intSecs);
}
for ($i=0; $i<$numBrwBuckets; $i++) {
$plot.=sprintf("$SEP%d", $lustreCltRpcWriteTot[$i]/$intSecs);
}
}
}
}
#ELAN
if ($subsys=~/x/ && $NumXRails)
{
$elanErrors=$elanSendFailTot+$elanNeterrAtomicTot+$elanNeterrDmaTot;
$plot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$elanRxTot/$intSecs, $elanTxTot/$intSecs,
$elanRxKBTot/$intSecs, $elanTxKBTot/$intSecs,
$elanErrors/$intSecs);
}
# INFINIBAND
# Now if 'x' specified and neither ELAN or IB, we still want to print all 0s so lets
# do it here (we could have done it in the ELAN routines is we wanted to).
if ($subsys=~/x/ && ($NumHCAs || ($NumHCAs==0 && $NumXRails==0)))
{
$plot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$ibRxTot/$intSecs, $ibTxTot/$intSecs,
$ibRxKBTot/$intSecs, $ibTxKBTot/$intSecs,
$ibErrorsTotTot);
}
# TCP
if ($subsys=~/t/)
{
# while tempted to control printing via $tcpFilt, by doing them all, we have a more
# consistent file that is easier to plot and not much more expensive in size
$plot.=sprintf("$SEP%$FS", $ipErrors/$intSecs);
$plot.=sprintf("$SEP%$FS", $tcpErrors/$intSecs);
$plot.=sprintf("$SEP%$FS", $udpErrors/$intSecs);
$plot.=sprintf("$SEP%$FS", $icmpErrors/$intSecs);
$plot.=sprintf("$SEP%$FS", $tcpData{TcpExt}->{TCPLoss}/$intSecs);
$plot.=sprintf("$SEP%$FS", $tcpData{TcpExt}->{TCPFastRetrans}/$intSecs);
}
# SLAB
if ($subsys=~/y/)
{
$plot.=sprintf("$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d",
$slabObjActTotal, $slabObjActTotalB, $slabObjAllTotal, $slabObjAllTotalB,
$slabSlabActTotal, $slabSlabActTotalB, $slabSlabAllTotal, $slabSlabAllTotalB,
$slabNumAct, $slabNumTot,6);
}
# BUDDYINFO
if ($subsys=~/b/)
{
for (my $i=0; $i<11; $i++)
{
$plot.=sprintf("$SEP%d", $buddyInfoTot[$i]);
}
}
# only if summary data
for (my $i=0; $impSummaryFlag && $i<$impNumMods; $i++)
{
&{$impPrintPlot[$i]}(3, \$plot) if $impOpts[$i]=~/s/;
}
writeData(0, $datetime, \$plot, $LOG, $ZLOG, 'log', \$oneline) if $netOpts!~/E/ || $netErrors;
}
###############################
# N O N - C O R E D A T A
###############################
if ($subsys=~/C/)
{
$cpuPlot='';
for ($i=0; $i<$NumCpus; $i++)
{
$cpuPlot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$userP[$i], $niceP[$i], $sysP[$i], $waitP[$i], $irqP[$i],
$softP[$i], $stealP[$i], $idleP[$i], $totlP[$i], $intrptTot[$i]/$intSecs);
}
writeData(0, $datetime, \$cpuPlot, CPU, $ZCPU, 'cpu', \$oneline);
}
#####################
# D S K F i l e
#####################
if ($subsys=~/D/)
{
$dskPlot='';
for (my $i=0; $i<@dskOrder; $i++)
{
$dskName=$dskOrder[$i];
next if ($dskFiltKeep eq '' && $dskName=~/$dskFiltIgnore/) || ($dskFiltKeep ne '' && $dskName!~/$dskFiltKeep/);
if (defined($disks{$dskName}))
{
my $i=$disks{$dskName};
$dskRecord=sprintf("%s$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$dskName,
$dskRead[$i]/$intSecs, $dskReadMrg[$i]/$intSecs, $dskReadKB[$i]/$intSecs,
$dskWrite[$i]/$intSecs, $dskWriteMrg[$i]/$intSecs, $dskWriteKB[$i]/$intSecs,
$dskRqst[$i], $dskQueLen[$i], $dskWait[$i], $dskSvcTime[$i], $dskUtil[$i]);
}
else
{
$dskRecord=sprintf("%s$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$dskName, 0,0,0,0,0,0,0,0,0,0,0);
}
# If exception processing in effect and writing to a file, make sure this entry
# qualities
if ($options=~/x/i)
{
# All we care about for I/O rates is if one is greater than exception.
$ios=$dskRead[$i]/$intSecs>=$limIOS || $dskWrite[$i]/$intSecs>=$limIOS;
$svc=$dskSvcTime[$i]*100;
# Either both tests are > limits or just one, depending on whether AND or OR
writeData(0, $datetime, \$dskRecord, DSKX, $ZDSKX, 'dskx', undef)
if ($limBool && $ios && $svc>=$limSVC) || (!$limBool && ($ios || $svc>=$limSVC));
}
# If not doing x-exception reporting, just build one long string
$dskPlot.="$SEP$dskRecord" if $options!~/x/;
}
# we only write DSK data when NOT doing x type execption processing
writeData(0, $datetime, \$dskPlot, DSK, $ZDSK, 'dsk', \$oneline) if $options!~/x/;
}
###############################
# E N V I R O N M E N T A L
###############################
if ($subsys=~/E/ && $interval3Print)
{
$envPlot='';
foreach $key (sort keys %$ipmiData)
{
for (my $i=0; $i{$key}}); $i++)
{
my $name= $ipmiData->{$key}->[$i]->{name};
my $inst= $ipmiData->{$key}->[$i]->{inst};
my $value= $ipmiData->{$key}->[$i]->{value};
my $status=$ipmiData->{$key}->[$i]->{status};
$value=0 if $value eq '';
$envPlot.="$SEP$value";
}
}
writeData(0, $datetime, \$envPlot, ENV, $ZENV, 'env', \$oneline);
}
##########################################
# L U S T R E D E T A I L F i l e
##########################################
if ($subsys=~/L/)
{
if ($reportOstFlag)
{
# Basic I/O always there and grouped together
$ostPlot='';
for ($i=0; $i<$NumOst; $i++)
{
$ostPlot.=sprintf("$SEP%s$SEP%d$SEP%d$SEP%d$SEP%d",
$lustreOsts[$i],
$lustreReadOps[$i]/$intSecs, $lustreReadKBytes[$i]/$intSecs,
$lustreWriteOps[$i]/$intSecs, $lustreWriteKBytes[$i]/$intSecs);
}
# These guys are optional and follow ALL the basic stuff
for ($i=0; $lustOpts=~/B/ && $i<$NumOst; $i++)
{
for ($j=0; $j<$numBrwBuckets; $j++)
{ $ostPlot.=sprintf("$SEP%d", $lustreBufRead[$i][$j]/$intSecs); }
for ($j=0; $j<$numBrwBuckets; $j++)
{ $ostPlot.=sprintf("$SEP%d", $lustreBufWrite[$i][$j]/$intSecs); }
}
writeData(0, $datetime, \$ostPlot, OST, $ZOST, 'ost', \$oneline);
}
if ($lustOpts=~/D/)
{
$blkPlot='';
for ($i=0; $i<$NumLusDisks; $i++)
{
$blkPlot.=sprintf("$SEP%s$SEP%d$SEP%d",
$LusDiskNames[$i],
$lusDiskReads[$i][$LusMaxIndex]/$intSecs,
$lusDiskReadB[$i][$LusMaxIndex]*0.5/$intSecs);
for ($j=0; $j<$LusMaxIndex; $j++)
{
$temp=(defined($lusDiskReads[$i][$j])) ? $lusDiskReads[$i][$j]/$intSecs : 0;
$blkPlot.=sprintf("$SEP%d", $temp);
}
$blkPlot.=sprintf("$SEP%d$SEP%d",
$lusDiskWrites[$i][$LusMaxIndex]/$intSecs,
$lusDiskWriteB[$i][$LusMaxIndex]*0.5/$intSecs);
for ($j=0; $j<$LusMaxIndex; $j++)
{
$temp=(defined($lusDiskWrites[$i][$j])) ? $lusDiskWrites[$i][$j]/$intSecs : 0;
$blkPlot.=sprintf("$SEP%d", $temp);
}
}
writeData(0, $datetime, \$blkPlot, BLK, $ZBLK, 'blk', \$online);
}
if ($reportCltFlag)
{
$cltPlot='';
if ($lustOpts=~/O/) # either OST details or FS details but not both
{
for ($i=0; $i<$NumLustreCltOsts; $i++)
{
# when lustre first starts up none of these have values
$cltPlot.=sprintf("$SEP%s$SEP%s$SEP%d$SEP%d$SEP%d$SEP%d",
$lustreCltOstFS[$i], $lustreCltOsts[$i],
defined($lustreCltLunRead[$i]) ? $lustreCltLunRead[$i]/$intSecs : 0,
defined($lustreCltLunReadKB[$i]) ? $lustreCltLunReadKB[$i]/$intSecs : 0,
defined($lustreCltLunWrite[$i]) ? $lustreCltLunWrite[$i]/$intSecs : 0,
defined($lustreCltLunWriteKB[$i]) ? $lustreCltLunWriteKB[$i]/$intSecs : 0);
}
for ($i=0; $lustOpts=~/B/ && $i<$NumLustreCltOsts; $i++)
{
for ($j=0; $j<$numBrwBuckets; $j++)
{
$cltPlot.=sprintf("$SEP%3d", $lustreCltRpcRead[$i][$j]/$intSecs);
}
for ($j=0; $j<$numBrwBuckets; $j++)
{
$cltPlot.=sprintf("$SEP%3d", $lustreCltRpcWrite[$i][$j]/$intSecs);
}
}
}
else # must be FS
{
for ($i=0; $i<$NumLustreFS; $i++)
{
$cltPlot.=sprintf("$SEP%s$SEP%d$SEP%d$SEP%d$SEP%d",
$lustreCltFS[$i],
$lustreCltRead[$i]/$intSecs, $lustreCltReadKB[$i]/$intSecs,
$lustreCltWrite[$i]/$intSecs, $lustreCltWriteKB[$i]/$intSecs);
}
for ($i=0; $lustOpts=~/M/ && $i<$NumLustreFS; $i++)
{
$cltPlot.=sprintf("$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d",
$lustreCltOpen[$i]/$intSecs, $lustreCltClose[$i]/$intSecs,
$lustreCltGetattr[$i]/$intSecs, $lustreCltSetattr[$i]/$intSecs,
$lustreCltSeek[$i]/$intSecs, $lustreCltFsync[$i]/$intSecs,
$lustreCltDirtyHits[$i]/$intSecs, $lustreCltDirtyMiss[$i]/$intSecs);
}
for ($i=0; $lustOpts=~/R/ && $i<$NumLustreFS; $i++)
{
$cltPlot.=sprintf("$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d",
$lustreCltRAPendingTot, $lustreCltRAHitsTot, $lustreCltRAMissesTot,
$lustreCltRANotConTot, $lustreCltRAMisWinTot, $lustreCltRAFalGrabTot,
$lustreCltRALckFailTot, $lustreCltRAReadDiscTot, $lustreCltRAZeroLenTot,
$lustreCltRAZeroWinTot, $lustreCltRA2EofTot, $lustreCltRAHitMaxTot,
$lustreCltRAWrongTot);
}
}
writeData(0, $datetime, \$cltPlot, CLT, $ZCLT, 'clt', \$oneline);
}
}
#########################
# N U M A F i l e
#########################
if ($subsys=~/M/)
{
my $numaPlot='';
for (my $i=0; $i<$CpuNodes; $i++)
{
# don't see how total can ever be 0, but let's be careful anyways
my $misses=$numaStat[$i]->{for}+$numaStat[$i]->{miss};
my $hitrate=($misses) ? $numaStat[$i]->{hits}/($numaStat[$i]->{hits}+$misses)*100 : 100;
$numaPlot.=sprintf("$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%.2f",
$numaMem[$i]->{used}, $numaMem[$i]->{free}, $numaMem[$i]->{slab},
$numaMem[$i]->{map}, $numaMem[$i]->{anon},
$numaMem[$i]->{inact}, $hitrate);
}
writeData(0, $datetime, \$numaPlot, NUMA, $ZNUMA, 'numa', \$oneline);
}
#####################
# N F S F i l e
#####################
if ($subsys=~/F/)
{
$nfsPlot='';
if ($nfs2CFlag)
{
$nfsPlot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$nfs2CRead/$intSecs, $nfs2CWrite/$intSecs, $nfs2CLookup/$intSecs, $nfs2CGetattr/$intSecs,
$nfs2CSetattr/$intSecs, $nfs2CReaddir/$intSecs, $nfs2CCreate/$intSecs, $nfs2CRemove/$intSecs,);
$nfsPlot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$nfs2CRename/$intSecs, $nfs2CLink/$intSecs, $nfs2CReadlink/$intSecs, $nfs2CNull/$intSecs,
$nfs2CSymlink/$intSecs, $nfs2CMkdir/$intSecs, $nfs2CRmdir/$intSecs, $nfs2CFsstat/$intSecs);
}
if ($nfs2SFlag)
{
$nfsPlot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$nfs2SRead/$intSecs, $nfs2SWrite/$intSecs, $nfs2SLookup/$intSecs, $nfs2SGetattr/$intSecs,
$nfs2SSetattr/$intSecs, $nfs2SReaddir/$intSecs, $nfs2SCreate/$intSecs, $nfs2SRemove/$intSecs);
$nfsPlot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$nfs2SRename/$intSecs, $nfs2SLink/$intSecs, $nfs2SReadlink/$intSecs, $nfs2SNull/$intSecs,
$nfs2SSymlink/$intSecs, $nfs2SMkdir/$intSecs, $nfs2SRmdir/$intSecs, $nfs2SFsstat/$intSecs);
}
if ($nfs3CFlag)
{
$nfsPlot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$nfs3CRead/$intSecs, $nfs3CWrite/$intSecs, $nfs3CCommit/$intSecs, $nfs3CLookup/$intSecs,
$nfs3CAccess/$intSecs, $nfs3CGetattr/$intSecs, $nfs3CSetattr/$intSecs, $nfs3CReaddir/$intSecs,
$nfs3CCreate/$intSecs, $nfs3CRemove/$intSecs, $nfs3CRename/$intSecs, $nfs3CLink/$intSecs);
$nfsPlot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$nfs3CReadlink/$intSecs, $nfs3CNull/$intSecs, $nfs3CSymlink/$intSecs, $nfs3CMkdir/$intSecs,
$nfs3CRmdir/$intSecs, $nfs3CFsstat/$intSecs, $nfs3CFsinfo/$intSecs, $nfs3CPathconf/$intSecs,
$nfs3CMknod/$intSecs, $nfs3CReaddirplus/$intSecs);
}
if ($nfs3SFlag)
{
$nfsPlot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$nfs3SRead/$intSecs, $nfs3SWrite/$intSecs, $nfs3SCommit/$intSecs, $nfs3SLookup/$intSecs,
$nfs3SAccess/$intSecs, $nfs3SGetattr/$intSecs, $nfs3SSetattr/$intSecs, $nfs3SReaddir/$intSecs,
$nfs3SCreate/$intSecs, $nfs3SRemove/$intSecs, $nfs3SRename/$intSecs, $nfs3SLink/$intSecs);
$nfsPlot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$nfs3SReadlink/$intSecs, $nfs3SNull/$intSecs, $nfs3SSymlink/$intSecs, $nfs3SMkdir/$intSecs,
$nfs3SRmdir/$intSecs, $nfs3SFsstat/$intSecs, $nfs3SFsinfo/$intSecs, $nfs3SPathconf/$intSecs,
$nfs3SMknod/$intSecs, $nfs3SReaddirplus/$intSecs);
}
if ($nfs4CFlag)
{
$nfsPlot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$nfs4CRead/$intSecs, $nfs4CWrite/$intSecs, $nfs4CCommit/$intSecs, $nfs4CLookup/$intSecs,
$nfs4CAccess/$intSecs, $nfs4CGetattr/$intSecs, $nfs4CSetattr/$intSecs, $nfs4CReaddir/$intSecs,
$nfs4CCreate/$intSecs, $nfs4CRemove/$intSecs, $nfs4CRename/$intSecs, $nfs4CLink/$intSecs);
$nfsPlot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$nfs4CReadlink/$intSecs, $nfs4CNull/$intSecs, $nfs4CSymlink/$intSecs, $nfs4CFsinfo/$intSecs,
$nfs4CPathconf/$intSecs);
}
if ($nfs4SFlag)
{
$nfsPlot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$nfs4SRead/$intSecs, $nfs4SWrite/$intSecs, $nfs4SCommit/$intSecs, $nfs4SLookup/$intSecs,
$nfs4SAccess/$intSecs, $nfs4SGetattr/$intSecs, $nfs4SSetattr/$intSecs, $nfs4SReaddir/$intSecs,
$nfs4SCreate/$intSecs, $nfs4SRemove/$intSecs, $nfs4SRename/$intSecs, $nfs4SLink/$intSecs,
$nfs4SReadlink/$intSecs);
}
writeData(0, $datetime, \$nfsPlot, NFS, $ZNFS, 'nfs', \$oneline);
}
#####################
# N E T F i l e
#####################
if ($subsys=~/N/)
{
$netPlot='';
for (my $i=0; $i<@netOrder; $i++)
{
# remember the order includes the speed
$netName=$netOrder[$i];
$netName=~s/:.*//;
next if ($netFiltKeep eq '' && $netName=~/$netFiltIgnore/) || ($netFiltKeep ne '' && $netName!~/$netFiltKeep/);
# remember 'err' is a single error counter and 'errs' is the total of those counters
# we also have to be sure to preseve network order
if (defined($networks{$netName}))
{
my $i=$networks{$netName};
$netPlot.=sprintf("$SEP%s$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$netName,
$netRxPkt[$i]/$intSecs, $netTxPkt[$i]/$intSecs,
$netRxKB[$i]/$intSecs, $netTxKB[$i]/$intSecs,
$netRxErr[$i]/$intSecs, $netRxDrp[$i]/$intSecs,
$netRxFifo[$i]/$intSecs,$netRxFra[$i]/$intSecs,
$netRxCmp[$i]/$intSecs, $netRxMlt[$i]/$intSecs,
$netTxErr[$i]/$intSecs, $netTxDrp[$i]/$intSecs,
$netTxFifo[$i]/$intSecs,$netTxColl[$i]/$intSecs,
$netTxCar[$i]/$intSecs, $netTxCmp[$i]/$intSecs,
$netRxErrs[$i]/$intSecs,$netTxErrs[$i]/$intSecs);
$netErrors+=$netRxErrs[$i]+$netTxErrs[$i];
}
else
{
$netPlot.=sprintf("$SEP%s$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS", $netName, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0);
}
}
# since we can't have holes in a line, with --netopts E we print ALL interfaces in the offending interval
writeData(0, $datetime, \$netPlot, NET, $ZNET, 'net', \$oneline) if $netOpts!~/E/ || $netErrors;
}
############################
# I n t e r c o n n e c t
############################
# Quadrics
if ($subsys=~/X/ && $NumXRails)
{
$elanPlot='';
for ($i=0; $i<$NumXRails; $i++)
{
$elanPlot.=sprintf("$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$elanRx[$i], $elanTx[$i], $elanRxKB[$i], $elanTxKB[$i],
$elanGet[$i], $elanPut[$i], $elanGetKB[$i], $elanPutKB[$i],
$elanComp[$i], $elanCompKB[$i],
$elanSendFail[$i], $elanNeterrAtomic[$i], $elanNeterrDma[$i]);
}
writeData(0, $datetime, \$elanPlot, ELN, $ZELN, 'eln', \$oneline);
}
# INFINIBAND
if ($subsys=~/X/ && $NumHCAs)
{
$ibPlot='';
for ($i=0; $i<$NumHCAs; $i++)
{
$ibPlot.=sprintf("$SEP%d$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS$SEP%$FS",
$i,
$ibRx[$i]/$intSecs, $ibTx[$i]/$intSecs,
$ibRxKB[$i]/$intSecs, $ibTxKB[$i]/$intSecs,
$ibErrorsTot[$i]);
}
writeData(0, $datetime, \$ibPlot, IB, $ZIB, 'ib', \$oneline);
}
#######################
# T C P F i l e
#######################
if ($subsys=~/T/)
{
# This is going to be big!!!
$tcpPlot='';
for my $type ('Ip', 'Tcp', 'Udp', 'Icmp', 'IpExt', 'TcpExt')
{
next if $type eq 'Ip' && $tcpFilt!~/i/;
next if $type eq 'Tcp' && $tcpFilt!~/t/;
next if $type eq 'Udp' && $tcpFilt!~/u/;
next if $type eq 'Icmp' && $tcpFilt!~/c/;
next if $type eq 'IpExt' && $tcpFilt!~/I/;
next if $type eq 'TcpExt' && $tcpFilt!~/T/;
# unfortunately the data is indexed by header name so we need an extra hop to get to it
foreach my $header (@{$tcpData{$type}->{hdr}})
{ $tcpPlot.=sprintf("$SEP%d", $tcpData{$type}->{$header}/$intSecs); }
}
writeData(0, $datetime, \$tcpPlot, TCP, $ZTCP, 'tcp', \$oneline);
}
#########################
# B U D D Y I N F O
#########################
if ($subsys=~/B/)
{
$budPlot='';
for (my $i=0; $i<$NumBud; $i++)
{
$budPlot.="$SEP$buddyNode[$i]$SEP$buddyZone[$i]";
for (my $j=0; $j<11; $j++)
{
$budPlot.=sprintf("$SEP%d", $buddyInfo[$i][$j]);
}
}
writeData(0, $datetime, \$budPlot, BUD, $ZBUD, 'bud', \$oneline);
}
#####################
# I M P O R T S
#####################
# only if detail data
for (my $i=0; $impDetailFlag && $i<$impNumMods; $i++)
{
if ($impOpts[$i]=~/d/)
{
$impPlot='';
&{$impPrintPlot[$i]}(4, \$impPlot);
if ($impPlot ne '')
{
$impDetFlag[$i]++;
writeData(0, $datetime, \$impPlot, , $impText[$i], $impGz[$i], $impKey[$i], \$oneline);
}
}
}
# F i n a l w r i t e
# we can't have holes in line so if --netopts E and no errors, do NOT print line to terminal or file.
return if $netOpts=~/E/ && !$netErrors;
# This write is necessary to write complete record to terminal or socket.
# Note if there is a socket error we're returning to the caller anyway;
writeData(1, $datetime, undef, $LOG, undef, undef, \$oneline)
if !$logToFileFlag || ($sockFlag && $export eq '');
}
# First and foremost, this is ONLY used to plot data. It will send it to the terminal,
# a socket, a data file or a combination of socket and data file.
# Secondly, we only call after processing a complete subsystem so in the case of
# core ones there's a single call but for detail subsystems one per.
# Therefore, when writing to a file, we write the whole string we're passed, but when
# writing to a terminal or socket, we build up one long string and write it on the
# last call. Since we can write to any combinations we need to handle them all.
sub writeData
{
my $eolFlag= shift;
my $datetime=shift;
my $string= shift;
my $file= shift;
my $zfile= shift;
my $errtxt= shift;
my $strall= shift;
# The very last call is special so handle it elsewhere
if (!$eolFlag)
{
# If writing to the terminal or a socket, just concatenate
# the strings together until the last call.
if (!$logToFileFlag || $sockFlag)
{
$$strall.=$$string;
}
# However, we might also be writing to a file as well as a socket
# and so need second test like this.
if ($logToFileFlag)
{
# Since we get called with !$eolFlag with partial lines, we always
# have a separator at the end of the line, so remove it before write.
my $localCopy=$$string;
$localCopy=~s/$SEP$//;
# Each record gets a timestamp and a newline. In the case of a file
# header, this will be null and the data will be the header!
$zfile->gzwrite("$datetime$localCopy\n") or
writeError($errtxt, $zfile) if $zFlag;
print {$file} "$datetime$localCopy\n" if !$zFlag;
}
return(1);
}
# Final Write!!!
# Doing these two writes this way will allow writing to the
# terminal AND a socket if we ever want to.
# NOTE - in virtually all cases there will be data to write. However when collecting
# data at different intervals, say with a custom import, there may not be data every
# time and we don't want to write empty records.
if ($$strall ne '')
{
if (!$sockFlag)
{
# final write to terminal
print "$datetime$$strall\n";
}
# write to socket but ONLY if we're not shutting down
if ($sockFlag && scalar(@sockets) && !$doneFlag)
{
# If a data line, preface with timestamp
$$strall="$datetime$$strall" if $strall!~/^#/;
# If we're not running a server, make sure each line begins
# with hostname and write to socket
$$strall=~s/^(.*)$/$Host $1/mg if !$serverFlag;
# we need to write to each listening socket, though there are probably rarely
# more than 1
$$strall.="\n";
foreach my $socket (@sockets)
{
my $length=length($$strall);
for (my $offset=0; $offset<$length;)
{
# Note - if there is a socket write error, writeData returns 0, but we're
# exiting this routine anyway and since '$doneFlag' is hopefully set because
# of a broken socket, the calling routines should exit cleanly.
# BUT only log error if in server mode, since normal as client
my $bytes=syswrite($socket, $$strall, $length, $offset);
if (!defined($bytes))
{
logmsg('E', "Error '$!' writing to socket") if $serverFlag;
return(0);
}
$offset+=$bytes;
$length-=$bytes;
}
}
}
}
return(1);
}
######################################
# T e r m i n a l F o r m a t s
######################################
sub printTerm
{
local $seconds=shift;
local $usecs= shift;
my ($ss, $mm, $hh, $mday, $mon, $year, $line, $i, $j);
# if someone wants to look at procs with --home and NOT --top, let them!
print "$clscr" if !$numTop && $homeFlag;
# There are a couple of things we want to do in interactive --top mode regardless
# of --brief or --verbose
if ($numTop && $playback eq '')
{
print $clscr if !$printTermFirst;
if ($printTermFirst) # --brief OR single subsys --verbose
{
# move the cursor to the correct location for ALL cases
if ($subsys ne 'Z')
{
my $lineNum=$totalCounter+2;
$lineNum=$scrollEnd if $lineNum>$scrollEnd;
$lineNum=0 if !$sameColsFlag || $detailFlag;
printf "%c[%d;H", 27, $lineNum;
}
# We only want to clear the screen once and print the header once
# the first time through and then just overpaint starting with data,
# unless of course we have details in which case we always print it.
$clscr=$home;
$headerRepeat=0 if !$detailFlag;
}
$printTermFirst=1;
}
# if we're including date and/or time, do once for whole interval
$line=$datetime='';
if ($miniDateFlag || $miniTimeFlag)
{
($ss, $mm, $hh, $mday, $mon, $year)=localtime($seconds);
$datetime=sprintf("%02d:%02d:%02d", $hh, $mm, $ss);
$datetime=sprintf("%02d/%02d %s", $mon+1, $mday, $datetime) if $options=~/d/;
$datetime=sprintf("%04d%02d%02d %s", $year+1900, $mon+1, $mday, $datetime) if $options=~/D/;
$datetime.=".$usecs" if ($options=~/m/);
$datetime.=" ";
}
################
# B r i e f
################
if ($briefFlag)
{
# This always goes to terminal or socket and is never compressed so we don't need
# all the options of writeData() [yet].
printBrief();
# --top mode requires process data too but only if interactive OR we're in playback
# mode and processing a file with process data in it
if ($numTop && ($playback eq '' || (($playback{$prefix}->{flags} & 1)==0) || $rawPFlag))
{
printTermProc() if $topProcFlag;
printTermSlab() if $topSlabFlag;
$headerRepeat=-1 && $playback eq ''; # only print header once interactively
}
return;
}
############################
# V e r b o s e
############################
# These interval counters will always match the interval we're about to print
$interval1Counter++ if $i1DataFlag;
$interval2Counter++ if $i2DataFlag && $interval2Print;
$interval3Counter++ if $i3DataFlag && $interval3Print;
# we usually want record break separators (with timestamps) except in a few cases which
$separatorHeaderPrinted=0;
if ($subsys=~/c/)
{
$i=$NumCpus;
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# CPU$Hyper SUMMARY (INTR, CTXSW & PROC $rate)$cpuDisabledMsg\n");
printText("#$miniDateTime User Nice Sys Wait IRQ Soft Steal Idle CPUs Intr Ctxsw Proc RunQ Run Avg1 Avg5 Avg15 RunT BlkT\n");
exit(0) if $showColFlag;
}
$line=sprintf("$datetime %4d %4d %4d %4d %4d %4d %4d %4d %4d %4s %4s %4d %4d %4d %5.2f %5.2f %5.2f %4d %4d\n",
$userP[$i], $niceP[$i], $sysP[$i], $waitP[$i],
$irqP[$i], $softP[$i], $stealP[$i], $idleP[$i],
$cpusEnabled,
cvt($intrpt/$intSecs), cvt($ctxt/$intSecs), $proc/$intSecs,
$loadQue, $loadRun, $loadAvg1, $loadAvg5, $loadAvg15, $procsRun, $procsBlock);
printText($line);
}
if ($subsys=~/C/)
{
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# SINGLE CPU$Hyper STATISTICS$cpuDisabledMsg\n");
my $intrptText=($subsys=~/j/i) ? ' INTRPT' : '';
printText("#$miniDateTime Cpu User Nice Sys Wait IRQ Soft Steal Idle$intrptText\n");
exit(0) if $showColFlag;
}
# if not recorded and user chose -s C don't print line items
if (defined($userP[0]))
{
for ($i=0; $i<$NumCpus; $i++)
{
# skip idle CPUs if --cpuopts z specified. I'm rather check for idle==100% but some kernels don't
# always increment counts and there are actually idle cpus with values of 0 here.
next if $cpuOpts=~/z/ && $userP[$i]+$niceP[$i]+$sysP[$i]+$waitP[$i]+$irqP[$i]+$softP[$i]+$stealP[$i]==0;
$line=sprintf("$datetime %4d %3d %3d %3d %3d %3d %3d %3d %3d",
$i,
$userP[$i], $niceP[$i], $sysP[$i], $waitP[$i],
$irqP[$i], $softP[$i], $stealP[$i], $idleP[$i]);
$line.=sprintf(" %6d", $intrptTot[$i]/$intSecs) if $subsys=~/j/i;
printText("$line\n");
}
}
}
# Only meaningful when Interrupts not combined with -sC
if ($subsys=~/j/ && !$CFlag)
{
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# INTERRUPT SUMMARY$cpuDisabledMsg\n");
my $oneline="#$miniDateTime ";
for (my $i=0; $i<$NumCpus; $i++)
{
my $cpuname=($cpuEnabled[$i]) ? "Cpu$i" : "CpuX";
$oneline.=sprintf(" %6s", $cpuname);
}
printText("$oneline\n");
exit(0) if $showColFlag;
}
my $oneline="$datetime ";
for (my $i=0; $i<$NumCpus; $i++)
{
$oneline.=sprintf(" %6d", $intrptTot[$i]);
}
printText("$oneline\n");
exit(0) if $showColFlag;
}
if ($subsys=~/J/)
{
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# INTERRUPT DETAILS$cpuDisabledMsg\n");
my $oneline="#$miniDateTime Int ";
for (my $i=0; $i<$NumCpus; $i++)
{
my $cpuname=($cpuEnabled[$i] || $subsys!~/c/i) ? "Cpu$i" : "CpuX";
$oneline.=sprintf(" %6s", $cpuname);
}
$oneline.=sprintf(" %-15s %s\n", 'Type', 'Device(s)');
printText($oneline);
exit(0) if $showColFlag;
}
foreach my $key (sort keys %intrptType)
{
my $linetot=0;
my $oneline="$datetime $key ";
for (my $i=0; $i<$NumCpus; $i++)
{
next if $key eq 'ERR' || $key eq 'MIS';
my $ints=($key=~/^\d/) ? $intrpt[$key]->[$i]/$intSecs : $intrpt{$key}->[$i]/$intSecs;
$oneline.=sprintf("%6d ", $ints);
$linetot+=$ints;
}
$oneline.=sprintf(" %s", $intName{$key}) if $key!~/ERR|MIS/;
printText("$oneline\n") if $linetot;
}
}
if ($subsys=~/d/)
{
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# DISK SUMMARY ($rate)\n");
printText("#${miniDateTime}KBRead RMerged Reads SizeKB KBWrite WMerged Writes SizeKB\n");
exit(0) if $showColFlag;
}
$line=sprintf("$datetime %6d %6d %6d %6d %6d %6d %6d %6d\n",
$dskReadKBTot/$intSecs, $dskReadMrgTot/$intSecs, $dskReadTot/$intSecs,
$dskReadTot ? $dskReadKBTot/$dskReadTot : 0,
$dskWriteKBTot/$intSecs, $dskWriteMrgTot/$intSecs, $dskWriteTot/$intSecs,
$dskWriteTot ? $dskWriteKBTot/$dskWriteTot : 0);
printText($line);
}
if ($subsys=~/D/)
{
# deal with --dskopts f format here
if (!defined($dskhdr1Format))
{
if ($dskOpts!~/f/)
{
$dskhdr1Format="<---------reads---------><---------writes---------><--------averages--------> Pct\n";
$dskhdr2Format=" KBytes Merged IOs Size KBytes Merged IOs Size RWSize QLen Wait SvcTim Util\n";
$dskdetFormat="%s%-11s %6d %6d %4s %4s %6d %6d %4s %4s %5d %5d %4d %4d %3d\n";
}
else
{
$dskhdr1Format="<---------reads----------><---------writes---------><---------averages----------> Pct\n";
$dskhdr2Format=" KBytes Merged IOs Size KBytes Merged IOs Size RWSize QLen Wait SvcTim Util\n";
$dskdetFormat="%s%-11s %7.1f %6.0f %4s %4s %7.1f %6.0f %4s %4s %6.1f %6.1f %6.1f %6.1f %3.0f\n";
}
}
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# DISK STATISTICS ($rate)\n");
printText("#$miniFiller $dskhdr1Format");
printText("#${miniDateTime}Name $dskhdr2Format");
exit(0) if $showColFlag;
}
for (my $o=0; $o<@dskOrder; $o++)
{
# preserve display order but skip any disks not seen this interval
$dskName=$dskOrder[$o];
$i=$disks{$dskName};
next if !defined($i);
next if ($dskFiltKeep eq '' && $dskName=~/$dskFiltIgnore/) || ($dskFiltKeep ne '' && $dskName!~/$dskFiltKeep/);
# Filter out lines of all zeros when requested
next if $dskOpts=~/z/ && ($dskReadKB[$i]+$dskReadMrg[$i]+$dskRead[$i]+
$dskWriteKB[$i]+$dskWriteMrg[$i]+$dskWrite[$i]+
$dskRqst[$i]+$dskQueLen[$i]+$dskWait[$i]+$dskSvcTime[$i]+$dskUtil[$i]==0);
# If exception processing in effect, make sure this entry qualities
next if $options=~/x/ && $dskRead[$i]/$intSecs<$limIOS && $dskWrite[$i]/$intSecs<$limIOS;
$line=sprintf($dskdetFormat,
$datetime, $dskName,
$dskReadKB[$i]/$intSecs, $dskReadMrg[$i]/$intSecs, cvt($dskRead[$i]/$intSecs),
$dskRead[$i] ? cvt($dskReadKB[$i]/$dskRead[$i],4,0,1) : 0,
$dskWriteKB[$i]/$intSecs, $dskWriteMrg[$i]/$intSecs, cvt($dskWrite[$i]/$intSecs),
$dskWrite[$i] ? cvt($dskWriteKB[$i]/$dskWrite[$i],4,0,1) : 0,
$dskRqst[$i], $dskQueLen[$i], $dskWait[$i], $dskSvcTime[$i], $dskUtil[$i]);
printText($line);
}
}
if ($subsys=~/f/)
{
if (printHeader())
{
my $temp=($nfsFilt ne '') ? "Filters: $nfsFilt" : '';
printText("\n") if !$homeFlag;
printText("# NFS SUMMARY ($rate) $temp\n");
$temp="#$miniFiller";
$temp.="<---------------------------server--------------------------->" if $nfsSFlag;
$temp.="<----------------client---------------->" if $nfsCFlag;
printText("$temp\n");
$temp="#$miniDateTime";
$temp.=" Reads Writes Meta Comm UDP TCP TCPConn BadAuth BadClnt " if $nfsSFlag;
$temp.=" Reads Writes Meta Comm Retrans Authref" if $nfsCFlag;
printText("$temp\n");
exit(0) if $showColFlag;
}
$line=$datetime;
$line.=sprintf(" %6s %6s %4s %4s %4s %4s %4s %4s %4s",
cvt($nfsSReadsTot/$intSecs,6), cvt($nfsSWritesTot/$intSecs,6),
cvt($nfsSMetaTot/$intSecs), cvt($nfsSCommitTot/$intSecs),
cvt($nfsUdpTot/$intSecs), cvt($nfsTcpTot/$intSecs),
cvt($nfsTcpConnTot/$intSecs), cvt($rpcBadAuthTot/$intSecs),
cvt($rpcBadClntTot/$intSecs))
if $nfsSFlag;
$line.=sprintf(" %6s %6s %4s %4s %4s %4s",
cvt($nfsCReadsTot/$intSecs,6), cvt($nfsCWritesTot/$intSecs,6),
cvt($nfsCMetaTot/$intSecs), cvt($nfsCCommitTot/$intSecs),
cvt($rpcRetransTot/$intSecs), cvt($rpcCredRefTot/$intSecs))
if $nfsCFlag;
$line.="\n";
printText($line);
}
if ($subsys=~/F/)
{
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# NFS SERVER/CLIENT DETAILS ($rate)\n");
# NOTE - we're not including V2 root/wrcache
printText("#${miniDateTime}Type Read Writ Comm Look Accs Gttr Sttr Rdir Cre8 Rmov Rnam Link Rlnk Null Syml Mkdr Rmdr Fsta Finf Path Mknd Rdr+\n");
exit(0) if $showColFlag;
}
# As an optimization, only show data where the filesystem is actually active but if --nfsopts z, only show
# entries with non-zero data. Currently only valid value for $nfsOpts is 'z'
if ($nfs2CFlag && $nfs2CSeen && ($nfsOpts!~/z/ || $nfs2CRead+$nfs2CWrite+$nfs2CMeta))
{
$line =sprintf("$datetime Clt2 %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s\n",
cvt($nfs2CRead/$intSecs), cvt($nfs2CWrite/$intSecs), '',
cvt($nfs2CLookup/$intSecs), '', cvt($nfs2CGetattr/$intSecs),
cvt($nfs2CSetattr/$intSecs), cvt($nfs2CReaddir/$intSecs), cvt($nfs2CCreate/$intSecs),
cvt($nfs2CRemove/$intSecs), cvt($nfs2CRename/$intSecs), cvt($nfs2CLink/$intSecs),
cvt($nfs2CReadlink/$intSecs),cvt($nfs2CNull/$intSecs), cvt($nfs2CSymlink/$intSecs),
cvt($nfs2CMkdir/$intSecs), cvt($nfs2CRmdir/$intSecs), cvt($nfs2CFsstat/$intSecs));
printText($line);
}
if ($nfs2SFlag && $nfs2SSeen && ($nfsOpts!~/z/ || $nfs2SRead+$nfs2SWrite+$nfs2SMeta))
{
$line =sprintf("$datetime Svr2 %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s\n",
cvt($nfs2SRead/$intSecs), cvt($nfs2SWrite/$intSecs), '',
cvt($nfs2SLookup/$intSecs), '', cvt($nfs2SGetattr/$intSecs),
cvt($nfs2SSetattr/$intSecs), cvt($nfs2SReaddir/$intSecs), cvt($nfs2SCreate/$intSecs),
cvt($nfs2SRemove/$intSecs), cvt($nfs2SRename/$intSecs), cvt($nfs2SLink/$intSecs),
cvt($nfs2SReadlink/$intSecs),cvt($nfs2SNull/$intSecs), cvt($nfs2SSymlink/$intSecs),
cvt($nfs2SMkdir/$intSecs), cvt($nfs2SRmdir/$intSecs), cvt($nfs2SFsstat/$intSecs));
printText($line);
}
if ($nfs3CFlag && $nfs3CSeen && ($nfsOpts!~/z/ || $nfs3CRead+$nfs3CWrite+$nfs3CMeta))
{
$line =sprintf("$datetime Clt3 %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s\n",
cvt($nfs3CRead/$intSecs), cvt($nfs3CWrite/$intSecs), cvt($nfs3CCommit/$intSecs),
cvt($nfs3CLookup/$intSecs), cvt($nfs3CAccess/$intSecs), cvt($nfs3CGetattr/$intSecs),
cvt($nfs3CSetattr/$intSecs), cvt($nfs3CReaddir/$intSecs), cvt($nfs3CCreate/$intSecs),
cvt($nfs3CRemove/$intSecs), cvt($nfs3CRename/$intSecs), cvt($nfs3CLink/$intSecs),
cvt($nfs3CReadlink/$intSecs),cvt($nfs3CNull/$intSecs), cvt($nfs3CSymlink/$intSecs),
cvt($nfs3CMkdir/$intSecs), cvt($nfs3CRmdir/$intSecs), cvt($nfs3CFsstat/$intSecs),
cvt($nfs3CFsinfo/$intSecs), cvt($nfs3CPathconf/$intSecs),cvt($nfs3CMknod/$intSecs),
cvt($nfs3CReaddirplus/$intSecs));
printText($line);
}
if ($nfs3SFlag && $nfs3SSeen && ($nfsOpts!~/z/ || $nfs3SRead+$nfs3SWrite+$nfs3SMeta))
{
$line =sprintf("$datetime Svr3 %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s\n",
cvt($nfs3SRead/$intSecs), cvt($nfs3SWrite/$intSecs), cvt($nfs3SCommit/$intSecs),
cvt($nfs3SLookup/$intSecs), cvt($nfs3SAccess/$intSecs), cvt($nfs3SGetattr/$intSecs),
cvt($nfs3SSetattr/$intSecs), cvt($nfs3SReaddir/$intSecs), cvt($nfs3SCreate/$intSecs),
cvt($nfs3SRemove/$intSecs), cvt($nfs3SRename/$intSecs), cvt($nfs3SLink/$intSecs),
cvt($nfs3SReadlink/$intSecs),cvt($nfs3SNull/$intSecs), cvt($nfs3SSymlink/$intSecs),
cvt($nfs3SMkdir/$intSecs), cvt($nfs3SRmdir/$intSecs), cvt($nfs3SFsstat/$intSecs),
cvt($nfs3SFsinfo/$intSecs), cvt($nfs3SPathconf/$intSecs),cvt($nfs3SMknod/$intSecs),
cvt($nfs3SReaddirplus/$intSecs));
printText($line);
}
# Not Used: Mkdir Mknod Readdirplus Fsstat Rmdir
if ($nfs4CFlag && $nfs4CSeen && ($nfsOpts!~/z/ || $nfs4CRead+$nfs4CWrite+$nfs4CMeta))
{
$line =sprintf("$datetime Clt4 %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s\n",
cvt($nfs4CRead/$intSecs), cvt($nfs4CWrite/$intSecs), cvt($nfs4CCommit/$intSecs),
cvt($nfs4CLookup/$intSecs), cvt($nfs4CAccess/$intSecs), cvt($nfs4CGetattr/$intSecs),
cvt($nfs4CSetattr/$intSecs), cvt($nfs4CReaddir/$intSecs), cvt($nfs4CCreate/$intSecs),
cvt($nfs4CRemove/$intSecs), cvt($nfs4CRename/$intSecs), cvt($nfs4CLink/$intSecs),
cvt($nfs4CReadlink/$intSecs),cvt($nfs4CNull/$intSecs), cvt($nfs4CSymlink/$intSecs),
'', '', '', cvt($nfs4CFsinfo/$intSecs), cvt($nfs4CPathconf/$intSecs));
printText($line);
}
if ($nfs4SFlag && $nfs4SSeen && ($nfsOpts!~/z/ || $nfs4SRead+$nfs4SWrite+$nfs4SMeta))
{
# Not Used: Null Pathconf Mkdir Mknod Readdirplus Fsinfo Fsstat Symlink Rmdir
$line =sprintf("$datetime Svr4 %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s %4s\n",
cvt($nfs4SRead/$intSecs), cvt($nfs4SWrite/$intSecs), cvt($nfs4SCommit/$intSecs),
cvt($nfs4SLookup/$intSecs), cvt($nfs4SAccess/$intSecs), cvt($nfs4SGetattr/$intSecs),
cvt($nfs4SSetattr/$intSecs), cvt($nfs4SReaddir/$intSecs), cvt($nfs4SCreate/$intSecs),
cvt($nfs4SRemove/$intSecs), cvt($nfs4SRename/$intSecs), cvt($nfs4SLink/$intSecs),
cvt($nfs4SReadlink/$intSecs));
printText($line);
}
}
if ($subsys=~/i/)
{
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# INODE SUMMARY\n");
printText("#${miniFiller} Dentries File Handles Inodes\n");
printText("#${miniDateTime} Number Unused Alloc % Max Number\n");
exit(0) if $showColFlag;
}
$line=sprintf("$datetime %7s %7s %6s %5.2f %6s\n",
cvt($dentryNum,7), cvt($dentryUnused,7),
cvt($filesAlloc,6), $filesMax ? $filesAlloc*100/$filesMax : 0,
cvt($inodeUsed,6));
printText($line);
}
# This is the normal output for an MDS and only skip if --lustopts D and only D
# noting D output (which itself is only for hp-sfs), is handled elsewhere
if ($subsys=~/l/ && $reportMdsFlag && $lustOpts ne 'D')
{
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# LUSTRE MDS SUMMARY ($rate)\n");
printText("#${miniDateTime} Getattr GttrLck StatFS Sync Gxattr Sxattr Connect Disconn");
printText(" Reint") if $cfsVersion lt '1.6.5';
printText(" Create Link Setattr Rename Unlink") if $cfsVersion ge '1.6.5';
printText("\n");
exit(0) if $showColFlag;
}
# Don't report if exception processing in effect and we're below limit
# NOTE - exception processing only for versions < 1.6.5
if ($options!~/x/ || $cfsVersion ge '1.6.5' || $lustreMdsReint/$intSecs>=$limLusReints)
{
$line.=sprintf("$datetime %7d %7d %7d %7d %7d %7d %7d %7d",
$lustreMdsGetattr/$intSecs, $lustreMdsGetattrLock/$intSecs,
$lustreMdsStatfs/$intSecs, $lustreMdsSync/$intSecs,
$lustreMdsGetxattr/$intSecs, $lustreMdsSetxattr/$intSecs,
$lustreMdsConnect/$intSecs, $lustreMdsDisconnect/$intSecs);
if ($cfsVersion lt '1.6.5')
{
$line.=sprintf(" %5d", $lustreMdsReint/$intSecs);
}
else
{
$line.=sprintf(" %6d %6d %7d %6d %6d",
$lustreMdsReintCreate/$intSecs, $lustreMdsReintLink/$intSecs,
$lustreMdsReintSetattr/$intSecs, $lustreMdsReintRename/$intSecs,
$lustreMdsReintUnlink/$intSecs);
}
}
$line.="\n";
printText($line);
}
# This is the normal output for an OST and only skip if --lustopts D and only D
# noting D output (which itself is only for hp-sfs), is handled elsewhere
if ($subsys=~/l/ && $reportOstFlag && $lustOpts ne 'D')
{
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# LUSTRE OST SUMMARY ($rate)\n");
if ($lustOpts!~/B/)
{
printText("#${miniDateTime} KBRead Reads SizeKB KBWrite Writes SizeKB\n");
}
else
{
printText("#${miniFiller}<----------------------reads-------------------------|");
printText("-----------------------writes------------------------->\n");
$temp='';
foreach my $i (@brwBuckets)
{ $temp.=sprintf(" %3dP", $i); }
printText("#${miniDateTime}RdK Rds$temp WrtK Wrts$temp\n");
}
exit(0) if $showColFlag;
}
$line=$datetime;
if ($lustOpts!~/B/)
{
$line.=sprintf(" %7d %6d %6s %7d %6d %6s",
$lustreReadKBytesTot/$intSecs, $lustreReadOpsTot/$intSecs,
$lustreReadOpsTot ? cvt($lustreReadKBytesTot/$lustreReadOpsTot,6,0,1) : 0,
$lustreWriteKBytesTot/$intSecs, $lustreWriteOpsTot/$intSecs,
$lustreWriteOpsTot ? cvt($lustreWriteKBytesTot/$lustreWriteOpsTot,6,0,1) : 0);
}
else
{
$line.=sprintf("%4s %4s",
cvt($lustreReadKBytesTot/$intSecs,4,0,1), cvt($lustreReadOpsTot/$intSecs));
for ($i=0; $i<$numBrwBuckets; $i++)
{
$line.=sprintf(" %4s", cvt($lustreBufReadTot[$i]/$intSecs));
}
$line.=sprintf(" %4s %4s",
cvt($lustreWriteKBytesTot/$intSecs,4,0,1), cvt($lustreWriteOpsTot/$intSecs));
for ($i=0; $i<$numBrwBuckets; $i++)
{
$line.=sprintf(" %4s", cvt($lustreBufWriteTot[$i]/$intSecs));
}
}
$line.="\n";
printText($line);
}
# NOTE - this only applies to hp-sfs
if ($subsys=~/l/ && ($reportMdsFlag || $reportOstFlag) && $lustOpts=~/D/)
{
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# LUSTRE DISK BLOCK LEVEL SUMMARY ($rate)\n#$miniFiller");
$temp='';
# not even room to preceed sizes with r/w's.
foreach my $i (@diskBuckets)
{
#last if $i>$LustreMaxBlkSize;
if ($i<1000) { $temp.=sprintf(" %3sK", $i) } else { $temp.=sprintf(" %3dM", $i/1024); }
}
printText("RdK Rds$temp WrtK Wrts$temp\n");
exit(0) if $showColFlag;
}
# Now do the data
$line=$datetime;
$line.=sprintf("%4s %4s",
cvt($lusDiskReadBTot[$LusMaxIndex]*0.5/$intSecs),
cvt($lusDiskReadsTot[$LusMaxIndex]/$intSecs));
for ($i=0; $i<$LusMaxIndex; $i++)
{
$line.=sprintf(" %4s", cvt($lusDiskReadsTot[$i]/$intSecs));
}
$line.=sprintf(" %4s %4s",
cvt($lusDiskWriteBTot[$LusMaxIndex]*0.5/$intSecs),
cvt($lusDiskWritesTot[$LusMaxIndex]/$intSecs));
for ($i=0; $i<$LusMaxIndex; $i++)
{
$line.=sprintf(" %4s", cvt($lusDiskWritesTot[$i]/$intSecs));
}
printText("$line\n");
}
if ($subsys=~/L/ && $reportOstFlag && ($lustOpts=~/B/ || $lustOpts!~/D/))
{
if (printHeader())
{
# build ost header, and when no date/time make it even 1 char less.
$temp="Ost". ' 'x$OstWidth;
$temp=substr($temp, 0, $OstWidth);
$temp=substr($temp, 0, $OstWidth-2).' ' if $miniFiller eq '';
# When doing dates/time shift first field over 1 to the left;
$fill1='';
if ($miniFiller ne '')
{
$fill1=substr($miniDateTime, 0, length($miniFiller)-1);
}
printText("\n") if !$homeFlag;
printText("# LUSTRE FILESYSTEM SINGLE OST STATISTICS ($rate)\n");
if ($lustOpts!~/B/)
{
printText("#$fill1$temp KBRead Reads SizeKB KBWrite Writes SizeKB\n");
}
else
{
$temp2='';
foreach my $i (@brwBuckets)
{ $temp2.=sprintf(" %3dP", $i); }
printText("#$fill1$temp RdK Rds$temp2 WrtK Wrts$temp2\n");
}
exit(0) if $showColFlag;
}
for ($i=0; $i<$NumOst; $i++)
{
# If exception processing in effect, make sure this entry qualities
next if $options=~/x/ &&
$lustreReadKBytes[$i]/$intSecs<$limLusKBS &&
$lustreWriteKBytes[$i]/$intSecs<$limLusKBS;
$line='';
if ($lustOpts!~/B/)
{
$line.=sprintf("$datetime%-${OstWidth}s %7d %6d %6d %7d %6d %6d\n",
$lustreOsts[$i],
$lustreReadKBytes[$i]/$intSecs, $lustreReadOps[$i]/$intSecs,
$lustreReadOps[$i] ? $lustreReadKBytes[$i]/$lustreReadOps[$i] : 0,
$lustreWriteKBytes[$i]/$intSecs, $lustreWriteOps[$i]/$intSecs,
$lustreWriteOps[$i] ? $lustreWriteKBytes[$i]/$lustreWriteOps[$i] : 0);
}
else
{
$line.=sprintf("$datetime%-${OstWidth}s %4s %4s",
$lustreOsts[$i],
cvt($lustreReadKBytes[$i]/$intSecs,4,0,1),
cvt($lustreReadOps[$i]/$intSecs));
for ($j=0; $j<$numBrwBuckets; $j++)
{
$line.=sprintf(" %4s", cvt($lustreBufRead[$i][$j]/$intSecs));
}
$line.=sprintf(" %4s %4s",
cvt($lustreWriteKBytes[$i]/$intSecs,4,0,1),
cvt($lustreWriteOps[$i]/$intSecs));
for ($j=0; $j<$numBrwBuckets; $j++)
{
$line.=sprintf(" %4s", cvt($lustreBufWrite[$i][$j]/$intSecs));
}
$line.="\n";
}
printText($line);
}
}
if ($subsys=~/L/ && $lustOpts=~/D/)
{
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# LUSTRE DISK BLOCK LEVEL DETAIL ($rate, units are 512 bytes)\n#$miniFiller");
$temp='';
foreach my $i (@diskBuckets)
{
#last if $i>$LustreMaxBlkSize;
if ($i<1000) { $temp.=sprintf(" %3sK", $i) } else { $temp.=sprintf(" %3dM", $i/1024); }
}
printText("DISK RdK Rds$temp WrtK Wrts$temp\n");
exit(0) if $showColFlag;
}
# Now do the data
for ($i=0; $i<$NumLusDisks; $i++)
{
$line=$datetime;
$line.=sprintf("%4s %4s %4s",
$LusDiskNames[$i],
cvt($lusDiskReadB[$i][$LusMaxIndex]*0.5/$intSecs),
cvt($lusDiskReads[$i][$LusMaxIndex]/$intSecs));
for ($j=0; $j<$LusMaxIndex; $j++)
{
$temp=(defined($lusDiskReads[$i][$j])) ? cvt($lusDiskReads[$i][$j]/$intSecs) : 0;
$line.=sprintf(" %4s", $temp);
}
$line.=sprintf(" %4s %4s",
cvt($lusDiskWriteB[$i][$LusMaxIndex]*0.5/$intSecs),
cvt($lusDiskWrites[$i][$LusMaxIndex]/$intSecs));
for ($j=0; $j<$LusMaxIndex; $j++)
{
$temp=(defined($lusDiskWrites[$i][$j])) ? cvt($lusDiskWrites[$i][$j]/$intSecs) : 0;
$line.=sprintf(" %4s", $temp);
}
printText("$line\n");
}
}
# NOTE - there are a number of different types of formats here and we're always going
# to include reads/writes with all of them!
if ($subsys=~/l/ && $reportCltFlag)
{
# If time for common header, do it...
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# LUSTRE CLIENT SUMMARY ($rate)");
printText(":") if $lustOpts=~/[BMR]/;
printText(" RPC-BUFFERS (pages)") if $lustOpts=~/B/;
printText(" METADATA") if $lustOpts=~/M/;
printText(" READAHEAD") if $lustOpts=~/R/;
printText("\n");
}
# If exception processing must be above minimum
if ($options!~/x/ ||
$lustreCltReadKBTot/$intSecs>=$limLusKBS ||
$lustreCltWriteKBTot/$intSecs>=$limLusKBS)
{
if ($lustOpts!~/[BMR]/)
{
printText("#$miniDateTime KBRead Reads SizeKB KBWrite Writes SizeKB\n")
if printHeader();
exit(0) if $showColFlag;
$line=sprintf("$datetime %7d %6d %6d %7d %6d %6d\n",
$lustreCltReadKBTot/$intSecs, $lustreCltReadTot/$intSecs,
$lustreCltReadTot ? int($lustreCltReadKBTot/$lustreCltReadTot) : 0,
$lustreCltWriteKBTot/$intSecs, $lustreCltWriteTot/$intSecs,
$lustreCltWriteTot ? int($lustreCltWriteKBTot/$lustreCltWriteTot) : 0);
printText($line);
}
if ($lustOpts=~/B/)
{
if (printHeader())
{
$temp='';
foreach my $i (@brwBuckets)
{ $temp.=sprintf(" %3dP", $i); }
printText("#${miniDateTime}RdK Rds$temp WrtK Wrts$temp\n");
exit(0) if $showColFlag;
}
$line="$datetime";
$line.=sprintf("%4s %4s",
cvt($lustreCltReadKBTot/$intSecs,4,0,1), cvt($lustreCltReadTot/$intSecs));
for ($i=0; $i<$numBrwBuckets; $i++)
{
$line.=sprintf(" %4s", cvt($lustreCltRpcReadTot[$i]/$intSecs));
}
$line.=sprintf(" %4s %4s",
cvt($lustreCltWriteKBTot/$intSecs,4,0,1), cvt($lustreCltWriteTot/$intSecs));
for ($i=0; $i<$numBrwBuckets; $i++)
{
$line.=sprintf(" %4s", cvt($lustreCltRpcWriteTot[$i]/$intSecs));
}
printText("$line\n");
}
if ($lustOpts=~/M/)
{
printText("#$miniDateTime KBRead Reads KBWrite Writes Open Close GAttr SAttr Seek Fsynk DrtHit DrtMis\n")
if printHeader();
exit(0) if $showColFlag;
$line=sprintf("$datetime %7d %6d %7d %6d %5d %5d %5d %5d %5d %5d %6d %6d\n",
$lustreCltReadKBTot/$intSecs, $lustreCltReadTot/$intSecs,
$lustreCltWriteKBTot/$intSecs, $lustreCltWriteTot/$intSecs,
$lustreCltOpenTot/$intSecs, $lustreCltCloseTot/$intSecs,
$lustreCltGetattrTot/$intSecs, $lustreCltSetattrTot/$intSecs,
$lustreCltSeekTot/$intSecs, $lustreCltFsyncTot/$intSecs,
$lustreCltDirtyHitsTot/$intSecs, $lustreCltDirtyMissTot/$intSecs);
printText($line);
}
if ($lustOpts=~/R/)
{
printText("#$miniDateTime KBRead Reads KBWrite Writes Pend Hits Misses NotCon MisWin FalGrb LckFal Discrd ZFile ZerWin RA2Eof HitMax Wrong\n")
if printHeader();
exit(0) if $showColFlag;
$line=sprintf("$datetime %7d %6d %7d %6d %5d %5d %6d %6d %6d %6d %6d %6d %6d %6d %6d %6d %6d\n",
$lustreCltReadKBTot/$intSecs, $lustreCltReadTot/$intSecs,
$lustreCltWriteKBTot/$intSecs, $lustreCltWriteTot/$intSecs,
$lustreCltRAPendingTot/$intSecs, $lustreCltRAHitsTot/$intSecs,
$lustreCltRAMissesTot/$intSecs, $lustreCltRANotConTot/$intSecs,
$lustreCltRAMisWinTot/$intSecs, $lustreCltRAFalGrabTot/$intSecs,
$lustreCltRALckFailTot/$intSecs, $lustreCltRAReadDiscTot/$intSecs,
$lustreCltRAZeroLenTot/$intSecs, $lustreCltRAZeroWinTot/$intSecs,
$lustreCltRA2EofTot/$intSecs, $lustreCltRAHitMaxTot/$intSecs,
$lustreCltRAWrongTot/$intSecs);
printText($line);
}
}
}
# NOTE -- there are 2 levels of details, both with and without --lustopts O
if ($subsys=~/L/ && $reportCltFlag)
{
if (printHeader())
{
# we need to build filesystem header, and when no date/time make it even 1
# char less.
$temp="Filsys". ' 'x$FSWidth;
$temp=substr($temp, 0, $FSWidth);
$temp=substr($temp, 0, $FSWidth-2).' ' if $miniFiller eq '';
# When doing dates/time, we also need to shift first field over 1 to the left;
$fill1='';
if ($miniFiller ne '')
{
$fill1=substr($miniDateTime, 0, length($miniFiller)-1);
}
printText("\n") if !$homeFlag;
printText("# LUSTRE CLIENT DETAIL ($rate)");
printText(":") if $lustOpts=~/[BMR]/;
printText(" RPC-BUFFERS (pages)") if $lustOpts=~/B/;
printText(" METADATA") if $lustOpts=~/M/;
printText(" READAHEAD") if $lustOpts=~/R/;
printText("\n");
}
if ($lustOpts=~/O/)
{
# Never for M or R
if ($lustOpts!~/B/)
{
$fill2=' 'x($OstWidth-3);
printText("#$fill1$temp Ost$fill2 KBRead Reads SizeKB KBWrite Writes SizeKB\n")
if printHeader();
exit(0) if $showColFlag;
for ($i=0; $i<$NumLustreCltOsts; $i++)
{
$line=sprintf("$datetime%-${FSWidth}s %-${OstWidth}s %7d %6d %6d %7d %6d %6d\n",
$lustreCltOstFS[$i], $lustreCltOsts[$i],
defined($lustreCltLunReadKB[$i]) ? $lustreCltLunReadKB[$i]/$intSecs : 0,
$lustreCltLunRead[$i]/$intSecs,
(defined($lustreCltLunReadKB[$i]) && $lustreCltLunRead[$i]) ? $lustreCltLunReadKB[$i]/$lustreCltLunRead[$i] : 0,
defined($lustreCltLunWriteKB[$i]) ? $lustreCltLunWriteKB[$i]/$intSecs : 0,
$lustreCltLunWrite[$i]/$intSecs,
(defined($lustreCltLunWriteKB[$i]) && $lustreCltLunWrite[$i]) ? $lustreCltLunWriteKB[$i]/$lustreCltLunWrite[$i] : 0);
printText($line);
}
}
if ($lustOpts=~/B/)
{
$fill2=' 'x($OstWidth-3);
if (printHeader())
{
$temp2=' 'x(length("$fill1$temp Ost$fill2 "));
$temp3='';
foreach my $i (@brwBuckets)
{ $temp3.=sprintf(" %3dP", $i); }
printText("#$fill1$temp Ost$fill2 RdK Rds$temp3 WrtK Wrts$temp3\n");
}
for ($clt=0; $clt<$NumLustreCltOsts; $clt++)
{
$line=sprintf("$datetime%-${FSWidth}s %-${OstWidth}s", $lustreCltOstFS[$clt], $lustreCltOsts[$clt]);
$line.=sprintf("%4s %4s",
cvt($lustreCltLunReadKB[$clt]/$intSecs,4,0,1), cvt($lustreCltLunRead[$clt]/$intSecs));
for ($i=0; $i<$numBrwBuckets; $i++)
{
$line.=sprintf(" %4s", cvt($lustreCltRpcRead[$clt][$i]/$intSecs));
}
$line.=sprintf(" %4s %4s",
cvt($lustreCltLunWriteKB[$clt]/$intSecs,4,0,1), cvt($lustreCltLunWrite[$clt]/$intSecs));
for ($i=0; $i<$numBrwBuckets; $i++)
{
$line.=sprintf(" %4s", cvt($lustreCltRpcWrite[$clt][$i]/$intSecs));
}
printText("$line\n");
}
}
}
else
{
$commonLine= "#$fill1$temp KBRead Reads SizeKB KBWrite Writes SizeKB";
if ($lustOpts!~/[MR]/)
{
printText("$commonLine\n") if printHeader();
exit(0) if $showColFlag;
for ($i=0; $i<$NumLustreFS; $i++)
{
$line=sprintf("$datetime%-${FSWidth}s %7d %6d %6d %7d %6d %6d\n",
$lustreCltFS[$i],
$lustreCltReadKB[$i]/$intSecs, $lustreCltRead[$i]/$intSecs,
$lustreCltRead[$i] ? $lustreCltReadKB[$i]/$lustreCltRead[$i] : 0,
$lustreCltWriteKB[$i]/$intSecs, $lustreCltWrite[$i]/$intSecs,
$lustreCltWrite[$i] ? $lustreCltWriteKB[$i]/$lustreCltWrite[$i] : 0);
printText($line);
}
}
if ($lustOpts=~/M/)
{
printText("$commonLine Open Close GAttr SAttr Seek Fsync DrtHit DrtMis\n")
if printHeader();
exit(0) if $showColFlag;
{
for ($i=0; $i<$NumLustreFS; $i++)
{
$line=sprintf("$datetime%-${FSWidth}s %7d %6d %6d %7d %6d %6d %5d %5d %5d %5d %5d %5d %6d %6d\n",
$lustreCltFS[$i],
$lustreCltReadKB[$i]/$intSecs, $lustreCltRead[$i]/$intSecs,
$lustreCltRead[$i] ? $lustreCltReadKB[$i]/$lustreCltRead[$i] : 0,
$lustreCltWriteKB[$i]/$intSecs, $lustreCltWrite[$i]/$intSecs,
$lustreCltWrite[$i] ? $lustreCltWriteKB[$i]/$lustreCltWrite[$i] : 0,
$lustreCltOpen[$i]/$intSecs, $lustreCltClose[$i]/$intSecs,
$lustreCltGetattr[$i]/$intSecs, $lustreCltSetattr[$i]/$intSecs,
$lustreCltSeek[$i]/$intSecs, $lustreCltFsync[$i]/$intSecs,
$lustreCltDirtyHits[$i]/$intSecs, $lustreCltDirtyMiss[$i]/$intSecs);
printText($line);
}
}
}
if ($lustOpts=~/R/)
{
printText("$commonLine Pend Hits Misses NotCon MisWin FalGrb LckFal Discrd ZFile ZerWin RA2Eof HitMax Wrong\n")
if printHeader();
exit(0) if $showColFlag;
{
for ($i=0; $i<$NumLustreFS; $i++)
{
$line=sprintf("$datetime%-${FSWidth}s %7d %6d %6d %7d %6d %6d %5d %5d %6d %6d %6d %6d %6d %6d %6d %6d %6d %6d %6d\n",
$lustreCltFS[$i],
$lustreCltReadKBTot/$intSecs, $lustreCltReadTot/$intSecs,
$lustreCltRead[$i] ? $lustreCltReadKB[$i]/$lustreCltRead[$i] : 0,
$lustreCltWriteKBTot/$intSecs, $lustreCltWriteTot/$intSecs,
$lustreCltWrite[$i] ? $lustreCltWriteKB[$i]/$lustreCltWrite[$i] : 0,
$lustreCltRAPendingTot/$intSecs, $lustreCltRAHitsTot/$intSecs,
$lustreCltRAMissesTot/$intSecs, $lustreCltRANotConTot/$intSecs, $lustreCltRAMisWinTot/$intSecs,
$lustreCltRAFalGrabTot/$intSecs, $lustreCltRALckFailTot/$intSecs, $lustreCltRAReadDiscTot/$intSecs,
$lustreCltRAZeroLenTot/$intSecs, $lustreCltRAZeroWinTot/$intSecs, $lustreCltRA2EofTot/$intSecs,
$lustreCltRAHitMaxTot/$intSecs, $lustreCltRAWrongTot/$intSecs);
printText($line);
}
}
}
}
}
if ($subsys=~/m/)
{
if (printHeader())
{
# Note that sar does page sizes in numbers of pages, not bytes
printText("\n") if !$homeFlag;
my $type=($memOpts!~/R/) ? '' : ' changes/int';
printText("# MEMORY SUMMARY$type\n");
if ($memOpts!~/R/)
{
$line="#$miniFiller";
$line.="<-------------------------------Physical Memory-------------------------------------->" if $memOpts eq '' || $memOpts=~/P/;
$line.="<-----------Swap------------><-------Paging------>" if $memOpts eq '' || $memOpts=~/V/;
$line.="<---Other---|-------Page Alloc------|------Page Refill----->" if $memOpts=~/p/;
$line.="<------Page Steal-------|-------Scan KSwap------|------Scan Direct----->" if $memOpts=~/s/;
printText("$line\n");
$line="#$miniFiller";
$line.=" Total Used Free Buff Cached Slab Mapped Anon Commit Locked Inact" if $memOpts eq '' || $memOpts=~/P/;
$line.=" Total Used Free In Out Fault MajFt In Out" if $memOpts eq '' || $memOpts=~/V/;
$line.=" Free Activ Dma Dma32 Norm Move Dma Dma32 Norm Move" if $memOpts=~/p/;
$line.=" Dma Dma32 Norm Move Dma Dma32 Norm Move Dma Dma32 Norm Move" if $memOpts=~/s/;
printText("$line\n");
}
else
{
$line=sprintf("#$miniFiller<-----------------------------------Physical Memory-------------------------------------------><------------Swap-------------><-------Paging------>\n");
printText($line);
printText("#$miniDateTime Total Used Free Buff Cached Slab Mapped Anon Commit Locked Inact Total Used Free In Out Fault MajFt In Out\n");
}
exit(0) if $showColFlag;
}
if ($memOpts!~/R/)
{
$line="$datetime ";
$line.=sprintf(" %7s %7s %7s %7s %7s %7s %7s %7s %7s %7s %5s",
cvt($memTot,7,1,1), cvt($memUsed,7,1,1), cvt($memFree,7,1,1),
cvt($memBuf,7,1,1), cvt($memCached,7,1,1),
cvt($memSlab,7,1,1), cvt($memMap,7,1,1), cvt($memAnon,7,1,1),
cvt($memCommit,7,1,1), cvt($memLocked,7,1,1), cvt($memInact,5,1,1)) if $memOpts eq '' || $memOpts=~/P/;
$line.=sprintf(" %5s %5s %5s %4s %4s %5s %5s %4s %4s",
cvt($swapTotal,5,1,1), cvt($swapUsed,5,1,1), cvt($swapFree,5,1,1),
cvt($swapin/$intSecs,5,1,1), cvt($swapout/$intSecs,5,1,1),
cvt($pagefault/$intSecs,5), cvt($pagemajfault/$intSecs,5),
cvt($pagein/$intSecs,4), cvt($pageout/$intSecs,4)) if $memOpts eq '' || $memOpts=~/V/;
$line.=sprintf(" %5s %5s %5s %5s %5s %5s %5s %5s %5s %5s",
cvt($pageFree,5), cvt($pageActivate,5),
cvt($pageAllocDma,5), cvt($pageAllocDma32,5), cvt($pageAllocNormal,5), cvt($pageAllocMove,5),
cvt($pageRefillDma,5), cvt($pageRefillDma32,5), cvt($pageRefillNormal,5), cvt($pageRefillMove,5)) if $memOpts=~/p/;
$line.=sprintf(" %5s %5s %5s %5s %5s %5s %5s %5s %5s %5s %5s %5s",
cvt($pageStealDma,5), cvt($pageStealDma32,5), cvt($pageStealNormal,5), cvt($pageStealMove,5),
cvt($pageKSwapDma,5), cvt($pageKSwapDma32,5), cvt($pageKSwapNormal,5), cvt($pageKSwapMove,5),
cvt($pageDirectDma,5), cvt($pageDirectDma32,5), cvt($pageDirectNormal,5), cvt($pageDirectMove,5)) if $memOpts=~/s/;
$line.="\n";
}
else
{
$line=sprintf("$datetime %7s %8s %8s %8s %8s %7s %7s %7s %7s %7s %6s %5s %6s %6s %4s %4s %5s %5s %4s %4s\n",
cvt($memTot/$intSecs,7,1,1), cvt($memUsedC/$intSecs,7,1,1), cvt($memFreeC/$intSecs,7,1,1),
cvt($memBufC/$intSecs,7,1,1), cvt($memCachedC/$intSecs,7,1,1),
cvt($memSlabC/$intSecs,7,1,1), cvt($memMapC/$intSecs,7,1,1), cvt($memAnonC/$intSecs,7,1,1),
cvt($memCommitC/$intSecs,7,1,1), cvt($memLockedC/$intSecs,7,1,1), cvt($memInactC/$intSecs,5,1,1),
cvt($swapTotal,5,1,1), cvt($swapUsedC/$intSecs,5,1,1), cvt($swapFreeC/$intSecs,5,1,1),
cvt($swapin/$intSecs,5,1,1), cvt($swapout/$intSecs,5,1,1),
cvt($pagefault/$intSecs,5), cvt($pagemajfault/$intSecs,5),
cvt($pagein/$intSecs,4), cvt($pageout/$intSecs,4));
}
printText($line);
}
if ($subsys=~/M/)
{
if (printHeader())
{
printText("\n") if !$homeFlag;
my $type=($memOpts!~/R/) ? '' : " change$type";
printText("# MEMORY STATISTICS $type\n");
{
# we've got the room so let's use an extra column for each and have the same
# headers for 'R' and because I'm lazy.
printText("#$miniFiller Node Total Used Free Slab Mapped Anon Locked Inact");
printText(" Hit%") if $memOpts!~/R/;
printText("\n");
}
exit(0) if $showColFlag;
}
$line='';
for (my $i=0; $i<$CpuNodes; $i++)
{
if ($memOpts!~/R/)
{
# total hits can be 0 if no data collected
my $misses=$numaStat[$i]->{for}+$numaStat[$i]->{miss};
my $hitrate=($misses) ? $numaStat[$i]->{hits}/($numaStat[$i]->{hits}+$misses)*100 : 100;
$line.=sprintf("$datetime %4d %8s %8s %8s %8s %8s %8s %8s %8s %6.2f\n", $i,
cvt($numaMem[$i]->{used}+$numaMem[$i]->{free},7,1,1),
cvt($numaMem[$i]->{used},7,1,1), cvt($numaMem[$i]->{free},7,1,1),
cvt($numaMem[$i]->{slab},7,1,1), cvt($numaMem[$i]->{map},7,1,1),
cvt($numaMem[$i]->{anon},7,1,1), cvt($numaMem[$i]->{lock},7,1,1),
cvt($numaMem[$i]->{inact},7,1,1), $hitrate);
}
else
{
$line.=sprintf("$datetime %4d %8s %8s %8s %8s %8s %8s %8s %8s\n", $i,
cvt($numaMem[$i]->{usedC}+$numaMem[$i]->{freeC},7,1,1),
cvt($numaMem[$i]->{usedC},7,1,1), cvt($numaMem[$i]->{freeC},7,1,1),
cvt($numaMem[$i]->{slabC},7,1,1), cvt($numaMem[$i]->{mapC},7,1,1),
cvt($numaMem[$i]->{anonC},7,1,1), cvt($numaMem[$i]->{lockC},7,1,1),
cvt($numaMem[$i]->{inactC},7,1,1));
}
}
printText($line);
}
if ($subsys=~/b/)
{
if (printHeader())
{
my $k=$PageSize/1024;
my $headers='';
for (my $i=0; $i<11; $i++)
{
my $header=sprintf("%dPg%s", 2**$i, $i==0 ? '': 's');
$headers.=sprintf("%8s", $header);
}
printText("\n") if !$homeFlag;
printText("# MEMORY FRAGMENTATION SUMMARY (${k}K pages)\n");
printText("#${miniDateTime}$headers\n");
exit(0) if $showColFlag;
}
my $line="$datetime ";
for (my $i=0; $i<11; $i++)
{
$line.=sprintf("%8d", $buddyInfoTot[$i]);
}
printText("$line\n");
}
if ($subsys=~/B/)
{
if (printHeader())
{
my $k=$PageSize/1024;
my $headers='';
for (my $i=0; $i<11; $i++)
{
my $header=sprintf("%dPg%s", 2**$i, $i==0 ? '': 's');
$headers.=sprintf("%8s", $header);
}
printText("\n") if !$homeFlag;
printText("# MEMORY FRAGMENTATION (${k}K pages)\n");
printText("#${miniDateTime}Node Zone $headers\n");
exit(0) if $showColFlag;
}
for (my $i=0; $i<$NumBud; $i++)
{
my $line="$datetime ";
$line.=sprintf("%4d %6s ", $buddyNode[$i], $buddyZone[$i]);
for (my $j=0; $j<11; $j++)
{
$line.=sprintf("%8d", $buddyInfo[$i][$j]);
}
printText("$line\n");
}
}
if ($subsys=~/n/)
{
my $netErrors=$netRxErrsTot+$netTxErrsTot;
if ($netOpts!~/E/ || $netErrors || $showColFlag)
{
if (printHeader())
{
my $errors=($netOpts=~/e/) ? 'ERRORS ' : '';
printText("\n") if !$homeFlag;
printText("# NETWORK ${errors}SUMMARY ($rate)\n");
printText("#${miniDateTime} KBIn PktIn SizeIn MultI CmpI ErrsI KBOut PktOut SizeO CmpO ErrsO\n")
if $netOpts!~/e/;
printText("#${miniDateTime} ErrIn DropIn FifoIn FrameIn ErrOut DropOut FifoOut CollOut CarrOut\n")
if $netOpts=~/e/;
exit(0) if $showColFlag;
}
# if --netopts E, only print lines when there are errors
# remember 'errs' is the totals of all the rx/tx counters, 'err' is a single counter
if ($netOpts!~/e/)
{
$line=sprintf("$datetime%6d %6d %6d %6d %6d %6d %6d %6d %6d %6d %6d\n",
$netRxKBTot/$intSecs, $netRxPktTot/$intSecs, $netRxPktTot ? $netRxKBTot*1024/$netRxPktTot : 0,
$netRxMltTot/$intSecs, $netRxCmpTot/$intSecs, $netRxErrsTot/$intSecs,
$netTxKBTot/$intSecs, $netTxPktTot/$intSecs, $netTxPktTot ? $netTxKBTot*1024/$netTxPktTot : 0,
$netTxCmpTot/$intSecs, $netTxErrsTot/$intSecs);
}
else
{
$line=sprintf("$datetime %7d %7d %7d %7d %7d %7d %7d %7d %7d\n",
$netRxErrTot/$intSecs, $netRxDrpTot/$intSecs, $netRxFifoTot/$intSecs, $netRxFraTot/$intSecs,
$netTxErrTot/$intSecs, $netTxErrTot/$intSecs, $netTxDrpTot/$intSecs, $netTxFifoTot/$intSecs,
$netTxCollTot/$intSecs, $netTxCarTot/$intSecs);
}
printText($line);
}
# When we skip printing an interval when a single subsystem, our header counter
# is off because it's been incremented, so back it up
elsif ($subsys eq 'n')
{
$interval1Counter--;
}
}
if ($subsys=~/N/)
{
# NOTE - header processing for detail data has always been ugly so let's not even
# deal with error exception processing.
if (printHeader())
{
my $errors=($netOpts=~/e/) ? 'ERRORS ' : '';
my $tempName=' 'x($NetWidth-5).'Name';
printText("\n") if !$homeFlag;
printText("# NETWORK ${errors}STATISTICS ($rate)\n");
printText("#${miniDateTime}Num $tempName KBIn PktIn SizeIn MultI CmpI ErrsI KBOut PktOut SizeO CmpO ErrsO\n")
if $netOpts!~/e/;
printText("#${miniDateTime}Num $tempName ErrIn DropIn FifoIn FrameIn ErrOut DropOut FifoOut CollOut CarrOut\n")
if $netOpts=~/e/;
exit(0) if $showColFlag;
}
for ($o=0; $o<@netOrder; $o++)
{
# since we want to print in the discovery order, we need to turn the next
# name into the right index
$netName=$netOrder[$o];
$i=$networks{$netName};
next if !defined($i);
next if ($netFiltKeep eq '' && $netName=~/$netFiltIgnore/) || ($netFiltKeep ne '' && $netName!~/$netFiltKeep/);
my $netErrors=$netRxErrs[$i]+$netTxErrs[$i];
if ($netOpts!~/e/)
{
$line=sprintf("$datetime %3d %${NetWidth}s %6d %6d %6d %6d %6d %6d %6d %6d %6d %6d %6d\n",
$i, $netName,
$netRxKB[$i]/$intSecs, $netRxPkt[$i]/$intSecs, $netRxPkt[$i] ? $netRxKB[$i]*1024/$netRxPkt[$i] : 0,
$netRxMlt[$i]/$intSecs, $netRxCmp[$i]/$intSecs, $netRxErrs[$i]/$intSecs,
$netTxKB[$i]/$intSecs, $netTxPkt[$i]/$intSecs, $netTxPkt[$i] ? $netTxKB[$i]*1024/$netTxPkt[$i] : 0,
$netTxCmp[$i]/$intSecs, $netTxErrs[$i]/$intSecs);
}
else
{
$line=sprintf("$datetime %3d %${NetWidth}s %7d %7d %7d %7d %7d %7d %7d %7d %7d\n",
$i, $netName[$i],
$netRxErr[$i]/$intSecs, $netRxDrp[$i]/$intSecs, $netRxFifo[$i]/$intSecs, $netRxFra[$i]/$intSecs,
$netTxErr[$i]/$intSecs, $netTxErr[$i]/$intSecs, $netTxDrp[$i]/$intSecs, $netTxFifo[$i]/$intSecs,
$netTxColl[$i]/$intSecs, $netTxCar[$i]/$intSecs);
}
printText($line) if $netOpts!~/E/ || $netErrors;
}
}
if ($subsys=~/s/)
{
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# SOCKET STATISTICS\n");
printText("#${miniFiller} <-------------Tcp-------------> Udp Raw <---Frag-->\n");
printText("#${miniDateTime}Used Inuse Orphan Tw Alloc Mem Inuse Inuse Inuse Mem\n");
exit(0) if $showColFlag;
}
$line=sprintf("$datetime%5d %5d %5d %5d %5d %5d %5d %5d %5d %5d\n",
$sockUsed, $sockTcp, $sockOrphan, $sockTw, $sockAlloc, $sockMem,
$sockUdp, $sockRaw, $sockFrag, $sockFragM);
printText($line);
}
if ($subsys=~/t/)
{
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# TCP STACK SUMMARY ($rate)\n");
$line= "#${miniFiller}";
$line.="<----------------------------------IpPkts----------------------------------->" if $tcpFilt=~/i/;
$line.="<---------------------------------Tcp--------------------------------->" if $tcpFilt=~/t/;
$line.="<------------Udp----------->" if $tcpFilt=~/u/;
$line.="<----------------------------Icmp--------------------------->" if $tcpFilt=~/c/;
$line.="<-------------------------IpExt------------------------>" if $tcpFilt=~/I/;
$line.="<------------------------------------------TcpExt----------------------------------------->" if $tcpFilt=~/T/;
$line.="\n";
$line.="#$miniFiller";
$line.=" Receiv Delivr Forwrd DiscdI InvAdd Sent DiscrO ReasRq ReasOK FragOK FragCr" if $tcpFilt=~/i/;
$line.=" ActOpn PasOpn Failed ResetR Estab SegIn SegOut SegRtn SegBad SegRes" if $tcpFilt=~/t/;
$line.=" InDgm OutDgm NoPort Errors" if $tcpFilt=~/u/;
$line.=" Recvd FailI UnreI EchoI ReplI Trans FailO UnreO EchoO ReplO" if $tcpFilt=~/c/;
$line.=" MPktsI BPktsI OctetI MOctsI BOctsI MPktsI OctetI MOctsI" if $tcpFilt=~/I/;
$line.=" FasTim Reject DelAck QikAck PktQue PreQuB HdPdct AkNoPy PreAck DsAcks RUData REClos SackS" if $tcpFilt=~/T/;
$line.="\n";
printText($line);
exit(0) if $showColFlag;
}
$line="$datetime ";
$line.=sprintf(" %6d %6d %6d %6d %6d %6d %6d %6d %6d %6d %6d",
$tcpData{Ip}->{InReceives}/$intSecs, $tcpData{Ip}->{InDelivers}/$intSecs,
$tcpData{Ip}->{ForwDatagrams}, $tcpData{Ip}->{InDiscards},
$tcpData{Ip}->{InAddrErrors}, $tcpData{Ip}->{OutRequests}/$intSecs,
$tcpData{Ip}->{OutDiscards}, $tcpData{Ip}->{ReasmReqds},
$tcpData{Ip}->{ReasmOKs}, $tcpData{Ip}->{FragOKs},
$tcpData{Ip}->{FragCreates})
if $tcpFilt=~/i/;
$line.=sprintf(" %6d %6d %6d %6d %6d %6d %6d %6d %6d %6d",
$tcpData{Tcp}->{ActiveOpens}/$intSecs, $tcpData{Tcp}->{PassiveOpens}/$intSecs,
$tcpData{Tcp}->{AttemptFails}, $tcpData{Tcp}->{EstabResets},
$tcpData{Tcp}->{CurrEstab}, $tcpData{Tcp}->{InSegs}/$intSecs,
$tcpData{Tcp}->{OutSegs}/$intSecs, $tcpData{Tcp}->{RetransSegs},
$tcpData{Tcp}->{InErrs}, $tcpData{Tcp}->{OutRsts})
if $tcpFilt=~/t/;
$line.=sprintf(" %6d %6d %6d %6d",
$tcpData{Udp}->{InDatagrams}/$intSecs, $tcpData{Udp}->{OutDatagrams}/$intSecs,
$tcpData{Udp}->{NoPorts}, $tcpData{Udp}->{InErrors})
if $tcpFilt=~/u/;
$line.=sprintf(" %5d %5d %5d %5d %5d %5d %5d %5d %5d %5d",
$tcpData{Icmp}->{InMsgs}, $tcpData{Icmp}->{InErrors},
$tcpData{Icmp}->{InDestUnreachs}, $tcpData{Icmp}->{InEchos},
$tcpData{Icmp}->{InEchoReps}, $tcpData{Icmp}->{OutMsgs},
$tcpData{Icmp}->{OutErrors}, $tcpData{Icmp}->{OutDestUnreachs},
$tcpData{Icmp}->{OutEchos}, $tcpData{Icmp}->{OutEchoReps})
if $tcpFilt=~/c/;
$line.=sprintf(" %6d %6d %6d %6d %6d %6d %6d %6d",
$tcpData{IpExt}->{InMcastPkts}, $tcpData{IpExt}->{InBcastPkts},
$tcpData{IpExt}->{InOctets}, $tcpData{IpExt}->{InMcastOctets},
$tcpData{IpExt}->{InBcastOctets}, $tcpData{IpExt}->{OutMcastPkts},
$tcpData{IpExt}->{OutOctets}, $tcpData{IpExt}->{OutMcastOctets})
if $tcpFilt=~/I/;
$line.=sprintf(" %6d %6d %6d %6d %6d %6d %6d %6d %6d %6d %6d %6d %6d",
$tcpData{TcpExt}->{TW}, $tcpData{TcpExt}->{PAWSEstab},
$tcpData{TcpExt}->{DelayedACKs}, $tcpData{TcpExt}->{DelayedACKLost},
$tcpData{TcpExt}->{TCPPrequeued}, $tcpData{TcpExt}->{TCPDirectCopyFromPrequeue},
$tcpData{TcpExt}->{TCPHPHits}, $tcpData{TcpExt}->{TCPPureAcks},
$tcpData{TcpExt}->{TCPHPAcks}, $tcpData{TcpExt}->{TCPDSACKOldSent},
$tcpData{TcpExt}->{TCPAbortOnData}, $tcpData{TcpExt}->{TCPAbortOnClose},
$tcpData{TcpExt}->{TCPSackShiftFallback})
if $tcpFilt=~/T/;
$line.="\n";
printText($line);
}
if ($subsys=~/E/ && $interval3Print)
{
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# ENVIRONMENTAL STATISTICS\n");
$envNewHeader=1;
}
my $keyCounter=0;
foreach $key (sort keys %$ipmiData)
{
next if $key=~/fan/ && $envOpts!~/f/;
next if $key=~/power/ && $envOpts!~/p/;
next if $key=~/temp/ && $envOpts!~/t/;
$keyCounter++;
if ($keyCounter==1 || $envOpts=~/M/)
{
$envHeader="#$miniDateTime";
$line="$datetime ";
}
for (my $i=0; $i{$key}}); $i++)
{
# we only do these when a main header printed
if ($envNewHeader)
{
my $name=$ipmiData->{$key}->[$i]->{name};
my $inst=$ipmiData->{$key}->[$i]->{inst};
$name=sprintf("$name%s", $inst ne '-1' ? $inst : '');
$envHeader.=sprintf(" %7s", $name);
}
# Not sure if I should be reporting 0 but that's why this is experimental!
my $value= $ipmiData->{$key}->[$i]->{value};
my $status=$ipmiData->{$key}->[$i]->{status};
$line.=sprintf(" %7s", ($value ne '') ? $value : 0);
}
# a multi-line print is done for each unique type (currently just fan & temp)
if ($envOpts=~/M/)
{
printText("$envHeader\n") if $envNewHeader;
printText("$line\n");
}
}
# Non-multi-line prints only done once
if ($envOpts!~/M/)
{
printText("$envHeader\n") if $envNewHeader;
exit(0) if $showColFlag;
printText("$line\n");
}
}
if ($subsys=~/x/)
{
if ($NumXRails)
{
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# ELAN4 SUMMARY ($rate)\n");
printText("#${miniDateTime}OpsIn OpsOut KBIn KBOut Errors\n");
exit(0) if $showColFlag;
}
$elanErrors=$elanSendFailTot+$elanNeterrAtomicTot+$elanNeterrDmaTot;
$line=sprintf("$datetime%6d %6d %6d %6d %6d\n",
$elanRxTot/$intSecs, $elanTxTot/$intSecs,
$elanRxKBTot/$intSecs, $elanTxKBTot/$intSecs,
$elanErrors/$intSecs);
printText($line);
}
if ($NumHCAs)
{
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# INFINIBAND SUMMARY ($rate)\n");
printText("#${miniDateTime} KBIn PktIn SizeIn KBOut PktOut SizeOut Errors\n");
exit(0) if $showColFlag;
}
$line=sprintf("$datetime%7d %7d %7d %7d %7d %7s %7s\n",
$ibRxKBTot/$intSecs, $ibRxTot/$intSecs, $ibRxTot ? cvt($ibRxKBTot*1024/$ibRxTot,7,0,1) : 0,
$ibTxKBTot/$intSecs, $ibTxTot/$intSecs, $ibTxTot ? cvt($ibTxKBTot*1024/$ibTxTot,7,0,1) : 0,
$ibErrorsTotTot);
printText($line);
}
}
if ($subsys=~/X/)
{
if ($NumXRails)
{
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# ELAN4 STATISTICS ($rate)\n");
printText("#${miniDateTime}Rail OpsIn OpsOut KB-In KB-Out OpsGet OpsPut KB-Get KB-Put Comp CompKB SndErr AtmErr DmsErr\n");
exit(0) if $showColFlag;
}
for ($i=0; $i<$NumXRails; $i++)
{
$line=sprintf("$datetime %4d %6d %6d %6d %6d %6d %6d %6d %6d %6d %6d %6d %6d %6d\n",
$i,
$elanRx[$i]/$intSecs, $elanTx[$i]/$intSecs,
$elanRxKB[$i]/$intSecs, $elanTxKB[$i]/$intSecs,
$elanGet[$i]/$intSecs, $elanPut[$i]/$intSecs,
$elanGetKB[$i]/$intSecs, $elanPutKB[$i]/$intSecs,
$elanComp[$i]/$intSecs, $elanCompKB[$i]/$intSecs,
$elanSendFail[$i]/$intSecs, $elanNeterrAtomic[$i]/$intSecs,
$elanNeterrDma[$i]/$intSecs);
printText($line);
}
}
if ($NumHCAs)
{
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# INFINIBAND STATISTICS ($rate)\n");
printText("#${miniDateTime}HCA KBIn PktIn SizeIn KBOut PktOut SizeOut Errors\n");
exit(0) if $showColFlag;
}
for ($i=0; $i<$NumHCAs; $i++)
{
$line=sprintf("$datetime %2d %7d %7d %7d %7d %7d %7d %7d\n",
$i,
$ibRxKB[$i]/$intSecs, $ibRx[$i]/$intSecs, $ibRx[$i] ? $ibRxKB[$i]/$ibRx[$i] : 0,
$ibTxKB[$i]/$intSecs, $ibTx[$i]/$intSecs, $ibTx[$i] ? $ibTxKB[$i]/$ibTx[$i] : 0,
$ibErrorsTot[$i]);
printText($line);
}
}
}
if ($subsys=~/y/ && $interval2Print)
{
if ($slabinfoFlag)
{
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# SLAB SUMMARY\n");
printText("#${miniFiller}<------------Objects------------><--------Slab Allocation-------><--Caches--->\n");
printText("#${miniDateTime} InUse Bytes Alloc Bytes InUse Bytes Total Bytes InUse Total\n");
exit(0) if $showColFlag;
}
$line=sprintf("$datetime %7s %7s %7s %7s %6s %7s %6s %7s %6s %6s\n",
cvt($slabObjActTotal,7), cvt($slabObjActTotalB,7,0,1),
cvt($slabObjAllTotal,7), cvt($slabObjAllTotalB,7,0,1),
cvt($slabSlabActTotal,6), cvt($slabSlabActTotalB,7,0,1),
cvt($slabSlabAllTotal,6), cvt($slabSlabAllTotalB,7,0,1),
cvt($slabNumAct,6), cvt($slabNumTot,6));
printText($line);
}
else
{
if (printHeader())
{
printText("\n") if !$homeFlag;
printText("# SLAB SUMMARY\n");
printText("#${miniFiller}<---Objects---><-Slabs-><-----memory----->\n");
printText("#${miniDateTime} In Use Avail Number Used Total\n");
exit(0) if $showColFlag;
}
$line=sprintf("$datetime %7s %7s %7s %7s %7s\n",
cvt($slabNumObjTot,7), cvt($slabObjAvailTot,7), cvt($slabNumTot,7),
cvt($slabUsedTot,7,0,1), cvt($slabTotalTot,7,0,1));
printText($line);
}
}
# tricky - by definitio --showcolheaders only shows single lines headers, SO if multiple
# imports and verbose, you only get the first!
for (my $i=0; $i<$impNumMods; $i++)
{
&{$impPrintVerbose[$i]}(printHeader(), $homeFlag, \$line);
printText($line) if $line ne ''; # rare, but it can happen when no instances of a component (screws up colmux!)
exit(0) if $showColFlag;
}
# Since slabs/processes both report rates, we need to skip first printable interval
# unless we're doing consecutive files
printTermSlab() if $subsys=~/Y/ && $interval2Print && (!$firstTime2 || $consecutiveFlag);
printTermProc() if $subsys=~/Z/ && $interval2Print && (!$firstTime2 || $consecutiveFlag);
# if running with --home in --top mode we might have junk in the rest of the display when
# items come and go, which they can when doing things like disk filtering or displaying
# processes so clear from the current location to the end of the display and reset $clscr
# so we never clear the screen more than once but rather just overwrite what's there
printText($clr) if $homeFlag || ($numTop && $playback eq '');
$clscr=$home;
}
sub printTermSlab
{
# Much of the top-slab methodology stolen from printTermProc()
my %slabSort;
my $slabCount=0;
my $eol=sprintf("%c[K", 27);
printf "%c[%d;H", 27, $scrollEnd ? $scrollEnd+1 : 0 if $numTop && $playback eq '';
# if someone wants to look at slabs with --home and NOT --top, let them!
print "$clscr" if !$numTop && $homeFlag;
if (printHeader() || $numTop)
{
if ($numTop)
{
$temp2=(split(/\s+/,localtime($seconds)))[3];
$temp2.=sprintf(".%03d", $usecs) if $options=~/m/;
}
printText("\n") if !$homeFlag;
my $temp=(!$topSlabFlag) ? 'SLAB DETAIL' : "TOP SLABS $temp2";
printText("# $temp\n");
if ($topSlabFlag)
{
print "#NumObj ActObj ObjSize NumSlab Obj/Slab TotSize TotChg TotPct Name\n";
}
elsif ($slabinfoFlag)
{
printText("#${miniFiller} <-----------Objects----------><---------Slab Allocation------><---Change-->\n");
printText("#${miniDateTime}Name InUse Bytes Alloc Bytes InUse Bytes Total Bytes Diff Pct\n");
}
else
{
printText("#${miniFiller} <----------- objects --------><--- slabs ---><---------allocated memory-------->\n");
printText("#${miniDateTime}Slab Name Size /slab In Use Avail SizeK Number UsedK TotalK Change Pct\n");
}
exit(0) if $showColFlag;
}
if ($slabinfoFlag)
{
for ($i=0; $i<$slabIndexNext; $i++)
{
if (!$topSlabFlag || $topType eq 'name') {
$key=$slabName[$i];
} elsif ($topType eq 'numobj') {
$key=sprintf('%9d', 999999999-$slabObjAllTot[$i]);
} elsif ($topType eq 'actobj') {
$key=sprintf('%9d', 999999999-$slabObjActTot[$i]);
} elsif ($topType eq 'objsize') {
$key=sprintf('%9d', 999999999-$slabObjSize[$i]);
} elsif ($topType eq 'numslab') {
$key=sprintf('%9d', 999999999-$slabSlabActTot[$i]);
} elsif ($topType eq 'objslab') {
$key=sprintf('%9d', 999999999-$slabObjPerSlab[$i]);
} elsif ($topType eq 'totsize') {
$key=sprintf('%9d', 999999999-$slabSlabAllTotB[$i]);
} elsif ($topType eq 'totchg') {
$key=sprintf('%9d', 999999999-abs($slabTotMemChg[$i]));
} elsif ($topType eq 'totpct') {
$key=sprintf('%9d', 999999999-abs($slabTotMemPct[$i]));
}
$slabSort{"$key-$i"}=$i; # need to include '-$i' to allow duplicates
}
foreach $key (sort keys %slabSort)
{
$i=$slabSort{$key};
# the first test is for filtering out zero-size slabs and the
# second for slabs that didn't change this during this interval
next if (($slabSlabAllTot[$i]==0 && ($topSlabFlag || $slabOpts=~/s/)) ||
($slabOpts=~/S/ && $slabSlabAct[$i]==0 && $slabSlabAll[$i]==0));
if ($topSlabFlag)
{
last if ++$slabCount>$numTop;
$line=sprintf("%7s %7s %7s %7s %7s %7s %7s %6.1f %s",
cvt($slabObjAllTot[$i],6), cvt($slabObjActTot[$i],6),
cvt($slabObjSize[$i],6), cvt($slabSlabActTot[$i],6),
cvt($slabObjPerSlab[$i],6), cvt($slabSlabAllTotB[$i],4,0,1),
cvt($slabTotMemChg[$i],4,0,1),$slabTotMemPct[$i], $slabName[$i]);
$line.=$eol if $playback eq '' && $numTop;
$line.="\n" if $playback ne '' || !$numTop || $slabCount<$numTop;
printText($line);
next;
}
$line=sprintf("$datetime%-25s %7s %7s %6s %7s %6s %7s %6s %7s %6s %6.1f\n",
substr($slabName[$i],0,25),
cvt($slabObjActTot[$i],6), cvt($slabObjActTotB[$i],7,0,1),
cvt($slabObjAllTot[$i],6), cvt($slabObjAllTotB[$i],7,0,1),
cvt($slabSlabActTot[$i],6), cvt($slabSlabActTotB[$i],7,0,1),
cvt($slabSlabAllTot[$i],6), cvt($slabSlabAllTotB[$i],7,0,1),
cvt($slabTotMemChg[$i],7,0,1),$slabTotMemPct[$i]);
printText($line);
}
}
else
{
foreach my $first (sort keys %slabfirst)
{
my $slab=$slabfirst{$first};
if (!$topSlabFlag || $topType eq 'name') {
$key=lc($first); # otherwise all upper-case names will come first
} elsif ($topType eq 'numobj') {
$key=sprintf('%9d', 999999999-$slabdata{$slab}->{slabsize}*$slabdata{$slab}->{avail});
} elsif ($topType eq 'actobj') {
$key=sprintf('%9d', 999999999-$slabdata{$slab}->{slabsize}*$slabdata{$slab}->{objects});
} elsif ($topType eq 'objsize') {
$key=sprintf('%9d', 999999999-$slabdata{$slab}->{slabsize});
} elsif ($topType eq 'numslab') {
$key=sprintf('%9d', 999999999-$slabdata{$slab}->{slabs});
} elsif ($topType eq 'objslab') {
$key=sprintf('%9d', 999999999-$slabdata{$slab}->{objper});
} elsif ($topType eq 'totsize') {
$key=sprintf('%9d', 999999999-$slabdata{$slab}->{total});
} elsif ($topType eq 'totchg') {
$key=sprintf('%9d', 999999999-abs($slabdata{$slab}->{memchg}));
} elsif ($topType eq 'totpct') {
$key=sprintf('%9d', 999999999-abs($slabdata{$slab}->{mempct}));
}
$slabSort{"$key-$first"}=$first; # need to include '-$first' to allow duplicates
}
foreach my $key (sort keys %slabSort)
{
my $first=$slabSort{$key};
my $slab=$slabfirst{$first};
# as for regular slabs, the first test is for filtering out zero-size
# slabs and the second for slabs that didn't change this during this interval
my $numObjects=$slabdata{$slab}->{objects};
my $numSlabs= $slabdata{$slab}->{slabs};
next if (($slabdata{$slab}->{objects}==0 && ($topSlabFlag || $slabOpts=~/s/)) ||
($slabOpts=~/S/ && $slabdata{$slab}->{lastobj}==$numObjects &&
$slabdata{$slab}->{lastslabs}==$numSlabs));
if ($topSlabFlag)
{
last if ++$slabCount>$numTop;
$line=sprintf("%7s %7s %7s %7s %7s %7s %7s %6.1f %s",
cvt($slabdata{$slab}->{slabsize}*$slabdata{$slab}->{avail},6),
cvt($slabdata{$slab}->{slabsize}*$numObjects,6),
cvt($slabdata{$slab}->{slabsize},6),
cvt($numSlabs,6),
cvt($slabdata{$slab}->{objper},6),
cvt($slabdata{$slab}->{total},4,0,1),
cvt($slabdata{$slab}->{memchg},4,0,1),
$slabdata{$slab}->{mempct}, $first);
$line.=$eol if $playback eq '' && $numTop;
$line.="\n" if $playback ne '' || !$numTop || $slabCount<$numTop;
printText($line);
next;
}
printf "$datetime%-25s %7d %5d %7d %7d %5d %7d %8d %8d %7s %6.1f\n",
substr($first,0,25),
$slabdata{$slab}->{slabsize},
$slabdata{$slab}->{objper},
$numObjects,
$slabdata{$slab}->{avail},
($PageSize<<$slabdata{$slab}->{order})/1024,
$numSlabs,
$slabdata{$slab}->{used}/1024,
$slabdata{$slab}->{total}/1024,
cvt($slabdata{$slab}->{memchg},7,0,1),
$slabdata{$slab}->{mempct};
# So we can tell when something changes
$slabdata{$slab}->{lastobj}= $numObjects;
$slabdata{$slab}->{lastslabs}=$numSlabs;
}
}
}
sub printTermProc
{
# if a process is discovered AFTER we start, this routine gets called called the first
# time a process is seen and '$interval2Secs' will be 0! In that one special case
# we need to wait for the next interval before printing.
return if !$interval2Secs;
# if we get here interactively, our cursor has already been set at home, but if
# --top and -s also specified ($scrollEnd!=0) we need to move past the scroll area
printf "%c[%d;H", 27, $scrollEnd ? $scrollEnd+1 : 0 if $numTop && $playback eq '';
# Never report timestamps in --top format.
my $tempFiller=(!$numTop) ? $miniDateTime : '';
my $tempTStamp=(!$numTop) ? $datetime : '';
# shorter name
$uw=$procUsrWidth;
# Since printHeader() is used by everyone, we need to force header printing for
# processes when in top mode since we ALWAYS want them
if (printHeader() || $numTop)
{
printText("\n") if !$homeFlag;
$temp1=($procOpts=~/f/) ? "(counters are cumulative)" : "(counters are $rate)";
$temp2='';
if ($numTop)
{
$temp2= " ".(split(/\s+/,localtime($seconds)))[3];
$temp2.=sprintf(".%03d", $usecs) if $options=~/m/;
printText("# TOP PROCESSES sorted by $topType $temp1$temp2\n");
}
else
{
printText("# PROCESS SUMMARY $temp1$temp2$cpuDisabledMsg\n");
}
$tempHdr='';
if ($procOpts!~/[im]/)
{
if ($procOpts!~/R/)
{
$prHeader='PR';
$prFormat='%2s';
}
else
{
$prHeader='PRIO';
$prFormat='%4d';
}
$tempHdr.="#${tempFiller} PID User $prHeader PPID THRD S VSZ RSS CP SysT UsrT Pct AccuTime ";
$tempHdr.=sprintf("%s ", $procOpts=~/s/ ? 'StrtTime' : 'StartTime ') if $procOpts=~/s/i;
$tempHdr.=" RKB WKB " if $processIOFlag;
$tempHdr.="VCtx NCtx " if $procOpts=~/x/;
$tempHdr.="MajF MinF Command\n";
}
elsif ($procOpts=~/i/)
{
$tempHdr.="#${tempFiller} PID User PPID S SysT UsrT Pct AccuTime RKB WKB RKBC WKBC RSys WSys Cncl Command\n";
}
elsif ($procOpts=~/m/)
{
$tempHdr.="#${tempFiller} PID User S VmSize VmLck VmRSS VmData VmStk VmExe VmLib VmSwp MajF MinF Command\n";
}
if ($procOpts=~/u/)
{
$user=sprintf("%-${uw}s", 'User');
$tempHdr=~s/User /$user/;
}
printText($tempHdr);
exit(0) if $showColFlag;
}
# When doing --top, we sort by time, io or faults
my %procSort;
my $eol='';
if ($numTop)
{
# clear from current position to the end of line since there could be junk there
$eol=sprintf("%c[K", 27) if $playback eq '';
foreach my $pid (keys %procIndexes)
{
# While I could do this at print time, it's more efficient to not even consider the
# during the sort.
next if $procState ne '' && $procState[$procIndexes{$pid}]!~/[$procState]/;
my $accum=0;
my $ipid=$procIndexes{$pid};
if ($topType eq 'vsz') {
$accum=defined($procVmSize[$ipid]) ? $procVmSize[$ipid] : 0;
} elsif ($topType eq 'rss') {
$accum=defined($procVmRSS[$ipid]) ? $procVmRSS[$ipid] : 0;
} elsif ($topType eq 'pid') {
$accum=32767-$pid; # to sort ascending
} elsif ($topType eq 'cpu') {
$accum=$NumCpus-$procCPU[$ipid]; # to sort ascending
} elsif ($topType eq 'syst') {
$accum=$procUTime[$ipid];
} elsif ($topType eq 'usrt') {
$accum=$procUTime[$ipid];
} elsif ($topType eq 'time') {
$accum=$procSTime[$ipid]+$procUTime[$ipid];
} elsif ($topType eq 'accum') {
$accum=$procSTimeTot[$ipid]+$procUTimeTot[$ipid];
} elsif ($topType eq 'thread') {
$accum=$procTCount[$ipid];
} elsif ($topType eq 'rkb') {
$accum=$procRKB[$ipid];
} elsif ($topType eq 'wkb') {
$accum=$procWKB[$ipid];
} elsif ($topType eq 'iokb') {
$accum=$procRKB[$ipid]+$procWKB[$ipid];
} elsif ($topType eq 'rbkc') {
$accum=$procRKBC[$ipid];
} elsif ($topType eq 'wkbc') {
$accum=$procWKBC[$ipid];
} elsif ($topType eq 'iokbc') {
$accum=$procRKBC[$ipid]+$procWKBC[$ipid];
} elsif ($topType eq 'ioall') {
$accum=$procRKB[$ipid]+ $procWKB[$ipid]+
$procRKBC[$ipid]+$procWKBC[$ipid];
} elsif ($topType eq 'rsys') {
$accum=$procRSys[$ipid];
} elsif ($topType eq 'wsys') {
$accum=$procWSys[$ipid];
} elsif ($topType eq 'iosys') {
$accum=$procRSys[$ipid]+$procWSys[$ipid];
} elsif ($topType eq 'iocncl') {
$accum=$procCKB[$ipid];
} elsif ($topType eq 'vctx') {
$accum=$procVCtx[$ipid];
} elsif ($topType eq 'nctx') {
$accum=$procNCtx[$ipid];
} elsif ($topType eq 'minf') {
$accum=$procMinFlt[$ipid];
} elsif ($topType eq 'flt') {
$accum=$procMajFlt[$ipid]+$procMinFlt[$ipid];
}
my $key=sprintf("%09d:%06d", 999999999-$accum, $pid);
$procSort{$key}=$pid if $procOpts!~/z/ || $accum!=0;
}
}
# otherwise we print in order of ascending pid
else
{
foreach $pid (keys %procIndexes)
{
next if $procState ne '' && $procState[$procIndexes{$pid}]!~/[$procState]/;
$procSort{sprintf("%06d", $pid)}=$pid;
}
}
my $procCount=0;
foreach $key (sort keys %procSort)
{
# if we had partial data for this pid don't try to print!
$i=$procIndexes{$procSort{$key}};
#print ">>>SKIP PRINTING DATA for pid $key i: $i"
# if (!defined($procSTimeTot[$i]));
next if (!defined($procSTimeTot[$i]));
last if $numTop && ++$procCount>$numTop;
# Handle -oF
if ($procOpts=~/f/)
{
$majFlt=$procMajFltTot[$i];
$minFlt=$procMinFltTot[$i];
}
else
{
$majFlt=$procMajFlt[$i]/$interval2Secs;
$minFlt=$procMinFlt[$i]/$interval2Secs;
}
# If wide mode OR when removing known shells (in which case we need to look at cmd1),
# we include the command arguments AND chop trailing spaces
($cmd0, $cmd1)=(defined($procCmd[$i])) ? split(/\s+/,$procCmd[$i],2) : ($procName[$i],'');
$cmd0=basename($cmd0) if $procOpts=~/r/ && $cmd0=~/^\//;
$cmd1='' if $procOpts!~/[kw]/ || !defined($cmd1);
# Since a program CAN modify its definition in /proc/pid/cmdline, it can
# end up without a trailing null and ultimately the split below results
# in an undefined $cmd1, which is why we need to test/init it if need be
if ($cmd1 ne '')
{
$cmd1=~s/\s+$//;
$cmd1=substr($cmd1, 0, $procCmdWidth);
}
# EXPERIMENTAL
# if told to do so, remove some of the known/standard shells from the command string in cmd0;
if ($procOpts=~/k/)
{
if ($cmd0=~m[/bin/sh|/usr/bin/perl|/usr/bin/python|^python])
{
$cmd1=~s/^-\S+\s+//; # remove optional switch some shells have
# now move 1st field in $cmd2 to $cmd1, which we no longer need, only keeping
# the cmd2 if procopts 'w' is set
$cmd1=~s/^(\S+)\s*//; # need '*' in case no args following command
$cmd0=$1;
}
$cmd1='' if $procOpts!~/w/;
}
# This is the standard format
if ($procOpts!~/[im]/)
{
# Note we only started fetching Tgid in V3.0.0
$line=sprintf("$tempTStamp%5d%s %-${uw}s $prFormat %5d %4d %1s %5s %5s %2d %s %s %s %s ",
$procPid[$i], $procThread[$i] ? '+' : ' ',
substr($procUser[$i],0,$uw), $procPri[$i],
defined($procTgid[$i]) && $procTgid[$i]!=$procPid[$i] ? $procTgid[$i] : $procPpid[$i],
$procTCount[$i], $procState[$i],
defined($procVmSize[$i]) ? cvt($procVmSize[$i],4,1,1) : 0,
defined($procVmRSS[$i]) ? cvt($procVmRSS[$i],4,1,1) : 0,
$procCPU[$i],
cvtT1($procSTime[$i]), cvtT1($procUTime[$i]),
cvtP($procSTime[$i]+$procUTime[$i]),
cvtT2($procSTimeTot[$i]+$procUTimeTot[$i]));
$line.=sprintf("%s ", cvtT5($procSTTime[$i])) if $procOpts=~/s/i;
$line.=sprintf("%4s %4s ",
cvt($procRKB[$i]/$interval2Secs,4,0,1),
cvt($procWKB[$i]/$interval2Secs,4,0,1)) if $processIOFlag;
$line.=sprintf("%4s %4s ",
cvt($procVCtx[$i]/$interval2Secs,4,0,1),
cvt($procNCtx[$i]/$interval2Secs,4,0,1)) if $procOpts=~/x/;
$line.=sprintf("%4s %4s %s %s",
cvt($majFlt), cvt($minFlt), $cmd0, $cmd1);
}
elsif ($procOpts=~/i/)
{
$line=sprintf("%s%5d%s %-${uw}s %5d %1s %s %s %3d %s ",
$tempTStamp, $procPid[$i], $procThread[$i] ? '+' : ' ',
substr($procUser[$i],0,$uw),
defined($procTgid[$i]) && $procTgid[$i]!=$procPid[$i] ? $procTgid[$i] : $procPpid[$i],
$procState[$i],
cvtT1($procSTime[$i]), cvtT1($procUTime[$i]),
cvtP($procSTime[$i]+$procUTime[$i]),
cvtT2($procSTimeTot[$i]+$procUTimeTot[$i]));
$line.=sprintf("%5s %5s %5s %5s %5s %5s %5s %s %s",
cvt($procRKB[$i]/$interval2Secs,5,0,1),
cvt($procWKB[$i]/$interval2Secs,5,0,1),
cvt($procRKBC[$i]/$interval2Secs,5,0,1),
cvt($procWKBC[$i]/$interval2Secs,5,0,1),
cvt($procRSys[$i]/$interval2Secs,5,0,1),
cvt($procWSys[$i]/$interval2Secs,5,0,1),
cvt($procCKB[$i]/$interval2Secs,5,0,1),
$cmd0, $cmd1);
}
elsif ($procOpts=~/m/)
{
$line=sprintf("%s%5d%s %-${uw}s %1s %6s %6s %6s %6s %6s %6s %6s %6s %4s %4s %s %s",
$tempTStamp, $procPid[$i], $procThread[$i] ? '+' : ' ',
substr($procUser[$i],0,$uw), $procState[$i],
defined($procVmSize[$i]) ? cvt($procVmSize[$i],6,1,1) : 0,
defined($procVmLck[$i]) ? cvt($procVmLck[$i],6,1,1) : 0,
defined($procVmRSS[$i]) ? cvt($procVmRSS[$i],6,1,1) : 0,
defined($procVmData[$i]) ? cvt($procVmData[$i],6,1,1) : 0,
defined($procVmStk[$i]) ? cvt($procVmStk[$i],6,1,1) : 0,
defined($procVmExe[$i]) ? cvt($procVmExe[$i],6,1,1) : 0,
defined($procVmLib[$i]) ? cvt($procVmLib[$i],6,1,1) : 0,
defined($procVmSwap[$i]) ? cvt($procVmSwap[$i],6,1,1) : 0,
cvt($majFlt), cvt($minFlt), $cmd0, $cmd1);
}
$line.=$eol if $playback eq '' && $numTop;
$line.="\n" if $playback ne '' || !$numTop || $procCount<$numTop;
printText($line);
}
# clear to the end of the display in case doing --procopts z, since the process list
# length changes dynamically
print $clr if $numTop && $playback eq '';
}
# this routine detects and 'fixes' counters that have wrapped
# *** warning *** It appears that partition 'use' counters wrap at wordsize/100
# on an ia32 (these are pretty pesky to actually catch). There may be more and
# they may behave differently on different architectures (though I tend to doubt
# it) so the best we can do is deal with them when we see them. It also looks like
# elan counters are divided by 1MB before reporting so we have to deal with them too
sub fix
{
my $counter=shift;
# if we're a smaller architecture than the number itself, we should still be
# ok because perl isn't restricted by word size.
if ($counter<0)
{
my $divisor=shift;
my $maxSize=shift;
# if param3 exists (rare), we use this as the max counter size; otherwidse 32 bit
my $wordsize=defined($maxSize) ? $maxSize : $word32;
# only adjust divisor when we're told to do so in param2.
my $add=defined($divisor) ? $wordsize/$divisor : $wordsize;
$counter+=$add;
}
return($counter);
}
# unitCounter 0 -> none, 1 -> K, etc (divide by $divisor this # times)
# divisor 0 -> /1000 1 -> /1024
sub cvt
{
my $field=shift;
my $width=shift;
my $unitCounter=shift;
my $divisorType=shift;
$width=4 if !defined($width);
$unitCounter=0 if !defined($unitCounter);
$divisorType=0 if !defined($divisorType);
$negative=0 if !defined($negative);
$field=int($field+.5) if $field>0; # round up in case <1
# This is tricky, because if the value fits within the width, we
# must also be sure the unit counter is 0 otherwise both may not
# fit. Naturally in 'wide' mode we aways report the complete value
# and we never print units with values of 0.
return($field) if ($field==0) || ($unitCounter<1 && length($field)<=$width) || $wideFlag;
# At least with slabs you can get negative numbers since we're tracking changes
my $sign=($field<0) ? -1 : 1;
$field=abs($field);
my $last=0;
my $divisor=($divisorType==0) ? 1000 : $OneKB;
while (length($field)>=$width)
{
$last=$field;
$field=int($field/$divisor);
$unitCounter++;
}
$field*=$sign;
my $units=substr(" KMGTP", $unitCounter, 1);
my $result=(abs($field)>0) ? "$field$units" : "1$units";
# Messy, but I hope reasonable efficient. We're only applying this to
# fields >= 'G' and options g/G! Furthermore, for -oG we only reformat
# when single digit because no room for 2.
if ($units=~/[GTP]/ && $options=~/g/i && (length($field))!=3)
{
# This one's a mouthful... we need to figure out what the remainer of the
# previous division was, by subtracting the field*divisor. Then we need
# to round up and pad with leading 0s. Note cases where we've rounded
# something like 9.999 which really needs to become 10.000
my $round=($options=~/g/) ? 5 : 50;
my $fraction=sprintf("%03d", $last-$field*$divisor+$round);
if ($fraction>=$divisor)
{
$field++;
$fraction='000';
}
# For 'G' we almost always print the first form but if we just rounded from 9.9 to
# to 10, we no longer have room for the fraction
if ($options=~/G/)
{
$result=(length($field)==1) ? "$field.".substr($fraction, 0, 1).'G' : "$field$units";
}
elsif ($options=~/g/)
{
# since the fraction follows the 'g', just chop the thing to 4 chars
$result=substr("${field}g".$fraction, 0, 4);
}
}
return($result);
}
# Time Format1 - convert time in jiffies to something ps-esque
# Seconds.hsec only (not suitable for longer times such as accumulated cpu)
sub cvtT1
{
my $jiffies=shift;
my $nsFlag= shift;
my ($secs, $hsec);
# set formatting for minutes according to 'no space' flag
$MF=(!$nsFlag) ? '%2d' : '%d';
$secs=int($jiffies/$HZ);
$jiffies=$jiffies-$secs*$HZ;
$hsec=$jiffies/$HZ*100;
return(sprintf("$MF.%02d", $secs, $hsec));
}
# Time Format1 - convert time in jiffies to something ps-esque
# we're not doing hours to save a couple of columns
sub cvtT2
{
my $jiffies=shift;
my $nsFlag= shift;
my ($hour, $mins, $secs, $time, $hsec);
$secs=int($jiffies/$HZ);
$jiffies=$jiffies-$secs*$HZ;
$hsec=$jiffies/$HZ*100;
$mins=int($secs/60);
$secs=$secs-$mins*60;
$time=($mins<60) ? sprintf("%02d:%02d.%02d", $mins % 60, $secs, $hsec) :
sprintf("%02d:%02d:%02d", int($mins/60), $mins % 60, $secs);
$time=" $time" if !$nsFlag && length($time)==8; # usually 8, but room for 3 digit mins
return($time);
}
sub cvtT3
{
my $secs=shift;
$secs/=100; # $secs really is msec
my $hours=int($secs/3600);
my $mins= int(($secs-$hours*3600)/60);
return(sprintf("%d:%02d:%02d", $hours, $mins, $secs-$hours*3600-$mins*60));
}
# convert time in seconds to date/time
sub cvtT4
{
my $seconds=shift;
my $msec=($options=~/m/) ? sprintf(".%s", (split(/\./, $seconds))[1]) : '';
my ($ss, $mm, $hh, $mday, $mon, $year)=localtime($seconds);
my $date=($options=~/d/) ?
sprintf("%02d/%02d", $mon+1, $mday) :
sprintf("%d%02d%02d", $year+1900, $mon+1, $mday);
my $time= sprintf("%02d:%02d:%02d%s", $hh, $mm, $ss, $msec);
return($date, $time);
}
sub cvtT5
{
my $time=shift;
my $realTime=$boottime+$time/100; # time in jiffies
my ($ss, $mm, $hh, $day, $mon)=localtime($realTime);
my $timestr;
if ($procOpts=~/s/)
{
$timestr=sprintf("%02d:%02d:%02d", $hh, $mm, $ss);
}
else
{
my $month=substr("JanFebMarAprMayJunJulAugSepOctNovDec", $mon*3, 3);
$timestr=sprintf("%s%02d-%02d:%02d:%02d", $month, $day, $hh, $mm, $ss);
}
return($timestr);
}
sub cvtP
{
my $jiffies=shift;
my ($secs, $percent);
# when using --from, we sometimes have not set $interval2SecsReal for the
# first sample so use i2 which is a good approximation
$secs=$jiffies/$HZ;
$interval2SecsReal=$interval2 if $interval2SecsReal==0;
$percent=sprintf("%3d", 100*$secs/$interval2SecsReal);
return($percent);
}
# Like printInterval, this is also used for terminal/socket output and therefore
# not something we need to worry about for logging!
sub printText
{
my $text=shift;
my $eol= shift;
print $text if !$sockFlag;
# just like in writeData, we need to make sure each line preceed
# with host name if not in server mode BUT only if not shutting down.
if ($sockFlag && scalar(@sockets) && !$doneFlag)
{
$text=~s/^(.*)$/$Host $1/mg if !$serverFlag;
$text.=">>><<<\n" if defined($eol);
foreach my $socket (@sockets)
{
my $length=length($text);
for (my $offset=0; $offset<$length;)
{
# When in client mode this WILL generate an error when the process who
# started us terminates.
my $bytes=syswrite($socket, $text, $length, $offset);
if (!defined($bytes))
{
logmsg('E', "Error '$!' writing to socket") if $serverFlag;
last;
}
$offset+=$bytes;
$length-=$bytes;
}
}
}
}
# see if time to print header
sub printHeader
{
# It might also be time to print a separator
printSeparator($seconds, $usecs) if !$separatorHeaderPrinted;
$separatorHeaderPrinted=1;
# S p e c i a l C a s e s
# Unless we say so explicitly we won't print a header and since we never do so under the
# following specific case of --top, let's get it out of the way first.
return(0) if $numTop && $headerRepeat==0 && $sameColsFlag;
return(1) if $subsys=~/[YZ]/ && $procFilt eq '' && $slabFilt eq '' && $slabOpts!~/S/;
return(1) if $numTop && $playback eq '';
return(1) if $headerRepeat==1; # brute force!
# S t a n d a r d P r o c e s s i n g
# The most common is when different column names and we simply do a new header every
# interval or when using --home to look top-ish output. Not sure why $totalCounter...
return(1) if $headerRepeat>-1 && (!$sameColsFlag || $totalCounter==1 || $homeFlag);
# Note that in detail mode (and that includes processes/slabs with filters) there's no
# real easy way to tell when to redo the header so rather we'll just repeat them every
# --hr set of intervals rather than lines.
return(1) if ($headerRepeat>0 &&
( ($interval1Counter % $headerRepeat)==1 ||
(($interval2Counter % $headerRepeat)==1 && $interval2Print) ||
(($interval3Counter % $headerRepeat)==1 && $interval3Print)) );
# do NOT print a header...
return(0);
}
# This routine gets called when it MIGHT be time to print a record separator since we've
# not printed one yet and are printing data for a new intercal.
sub printSeparator
{
my $seconds=shift;
my $usecs= shift;
# here's where we decide whether or not we really want the interval headers. This is also
# where all the special cases come in.
return if !$numTop && $sameColsFlag && $subsys!~/[YZ]/ && !$homeFlag;
return if $numTop && $playback eq '' && !$detailFlag;
return if $subsys eq 'Y' && ($slabFilt ne '' || $slabOpts=~/S/);
return if $subsys eq 'Z' && $procFilt ne '';
my $date=localtime($seconds);
if ($options=~/m/)
{
my ($dow, $mon, $day, $time, $year)=split(/ /, $date);
$date="$dow $mon $day $time.$usecs $year";
}
# Remember that -A with logging never writes to terminals.
my $temp=sprintf("%s", $homeFlag ? $clscr : "\n");
$temp.=sprintf("### RECORD %4d >>> $HostLC <<< ($seconds) ($date) ###\n", ++$separatorCounter);
printText($temp);
}
sub getHeader
{
my $file=shift;
my ($gzFlag, $header, $TEMP, $line);
$gzFlag=$file=~/gz$/ ? 1 : 0;
if ($gzFlag)
{
$TEMP=Compress::Zlib::gzopen($file, "rb") or logmsg("F", "Couldn't open '$file'");
}
else
{
open TEMP, "<$file" or logmsg("F", "Couldn't open '$file'");
}
$header="";
while (1)
{
$TEMP->gzreadline($line) if $gzFlag;
$line= if !$gzFlag;
last if $line!~/^#/;
$header.=$line;
}
close TEMP;
print "*** Header For: $file ***\n$header" if $debug & 16;
return($header);
}
sub incomplete
{
my $type=shift;
my $secs=shift;
my $special=shift;
my ($seconds, $ss, $mm, $hh, $mday, $mon, $year, $date, $time);
$seconds=(split(/\./, $secs))[0];
($ss, $mm, $hh, $mday, $mon, $year)=localtime($seconds);
$date=sprintf("%d%02d%02d", $year+1900, $mon+1, $mday);
$time=sprintf("%02d:%02d:%02d", $hh, $mm, $ss);
my $message=(!defined($special)) ? "Incomplete" : $special;
my $where=($playback eq '') ? "on $date" : "in $playbackFile";
logmsg("W", "$message data record skipped for $type data $where at $time");
}
# Handy for debugging
sub getTime
{
my $seconds=shift;
my ($ss, $mm, $hh, $mday, $mon, $year);
($ss, $mm, $hh, $mday, $mon, $year)=localtime($seconds);
return(sprintf("%02d:%02d:%02d", $hh, $mm, $ss));
}
########################################
# Brief Mode is VERY Special
########################################
sub printBrief
{
my ($command, $pad, $i);
my $line='';
# We want to track elapsed time. This is only looked at in interactive mode.
$miniStart=$seconds if !defined($miniStart) || $miniStart==0;
if ( $headerRepeat==1 ||
($headerRepeat==0 && !$headersPrinted) ||
($headerRepeat>0 && ($totalCounter % $headerRepeat)==1))
{
$cpuDisabledMsg=~s/^://; # just in case non-null
$pad=' ' x length($miniDateTime);
$fill1=($Hyper eq '') ? "----" : "";
$fill2=($Hyper eq '') ? "----" : "-";
$line.="$clscr";
$line.="#$cpuDisabledMsg\n" if $cpuDisabledMsg ne '';
$line.="#$pad";
$line.="<----${fill1}CPU$Hyper$fill2---->" if $subsys=~/c/;
if ($subsys=~/j/)
{
my $num=int(($NumCpus-1)*5/2);
my $pad1='-'x$num;
my $pad2=$pad1;
$line.="<${pad1}Int$pad2->";
}
# sooo ugly...
my ($tcp1,$tcp2);
if ($subsys=~/t/)
{
$tcp2='';
$tcp2.=' IP ' if $tcpFilt=~/i/;
$tcp2.=' Tcp ' if $tcpFilt=~/t/;
$tcp2.=' Udp ' if $tcpFilt=~/u/;
$tcp2.='Icmp ' if $tcpFilt=~/c/;
$tcp2.='TcpX ' if $tcpFilt=~/T/;
my $num=int((length($tcp2)-5)/2);
my $num2=((length($tcp2) % 2)==0) ? $num+1 : $num;
my $pre= '-' x $num;
my $post='-' x $num2;
$tcp1="<${pre}TCP$post>";
$tcp1="" if length($tcp2)==5;
}
$line.="<--Memory-->" if $subsys!~/m/ && $subsys=~/b/;
if ($memOpts!~/R/)
{
$line.="<-----------Memory----------->" if $subsys=~/m/ && $subsys!~/b/;
$line.="<-----------------Memory----------------->" if $subsys=~/m/ && $subsys=~/b/;
}
else
{
$line.="<--------------Memory-------------->" if $subsys=~/m/ && $subsys!~/b/;
$line.="<--------------------Memory-------------------->" if $subsys=~/m/ && $subsys=~/b/;
}
$line.="<-----slab---->" if $subsys=~/y/;
$line.="<----------Disks----------->" if $subsys=~/d/ && !$ioSizeFlag && $dskOpts!~/i/;
$line.="<---------------Disks---------------->" if $subsys=~/d/ && ($ioSizeFlag || $dskOpts=~/i/);
$line.="<----------Network---------->" if $subsys=~/n/ && !$ioSizeFlag && $netOpts!~/i/;
$line.="<---------------Network--------------->" if $subsys=~/n/ && ($ioSizeFlag || $netOpts=~/i/);
$line.=$tcp1 if $subsys=~/t/;
$line.="<------Sockets----->" if $subsys=~/s/;
$line.="<----Files--->" if $subsys=~/i/;
$line.="<---------------Elan------------->" if $subsys=~/x/ && $NumXRails;
$line.="<-----------InfiniBand----------->" if $subsys=~/x/ && ($NumHCAs || $NumHCAs+$NumXRails==0) && (!$ioSizeFlag && $xOpts!~/i/);
$line.="<----------------InfiniBand---------------->" if $subsys=~/x/ && ($NumHCAs || $NumHCAs+$NumXRails==0) && ($ioSizeFlag || $xOpts=~/i/);
# probably a better way to handle iosize too
$line=~s/Network/---Network---/ if $netOpts=~/e/;
# a bunch of extra work but worth it!
if ($subsys=~/f/)
{
# If all filters specified, no room!
if ($nfsFilt eq '' || length($nfsFilt)==17)
{
$line.="<------NFS Totals------>";
}
else
{
my $padL=$padR=int((14-length($nfsFilt))/2);
$padL++ if length($nfsFilt) & 1; # handle odd number of -'s
$padL='-'x$padL;
$padR='-'x$padR;
$line.="<$padL-NFS [$nfsFilt]-$padR>";
}
}
$line.="<--------Lustre MDS-------->" if $subsys=~/l/ && $reportMdsFlag;
$line.="<---------Lustre OST--------->" if $subsys=~/l/ && $reportOstFlag && !$ioSizeFlag;
$line.="<--------------Lustre OST-------------->" if $subsys=~/l/ && $reportOstFlag && $ioSizeFlag;
if ($subsys=~/l/ && $reportCltFlag)
{
$line.="<--------Lustre Client-------->" if !$ioSizeFlag && $lustOpts!~/R/;
$line.="<---------------Lustre Client--------------->" if !$ioSizeFlag && $lustOpts=~/R/;
$line.="<-------------Lustre Client------------->" if $ioSizeFlag && $lustOpts!~/R/;
$line.="<--------------------Lustre Client-------------------->" if $ioSizeFlag && $lustOpts=~/R/;
}
for (my $i=0; $i<$impNumMods; $i++) { &{$impPrintBrief[$i]}(1, \$line); }
$line.="\n";
$line.="#$miniDateTime";
$line.="cpu sys inter ctxsw " if $subsys=~/c/;
if ($subsys=~/j/)
{
# If < 10 cpus, use header of 'Cpu'. otherwiswe use 'Cp', 'C' or just the number.
# Naturally if more than a couple of dozen we'll need a very wide monitor.
$line.=sprintf("Cpu%d "x($NumCpus>10?10:$NumCpus), 0..$NumCpus);
$line.=sprintf("Cp%d "x($NumCpus>100?90:$NumCpus-10), 10..$NumCpus);
$line.=sprintf("C%d "x($NumCpus>1000?900:$NumCpus-100), 100..$NumCpus);
$line.=sprintf("%d "x($NumCpus-1000), 1000..$NumCpus);
# Rare, but if a cpu is offline, change its name in the header
if ($cpusDisabled)
{
for (my $i=0; $i<$NumCpus; $i++)
{
$line=~s/Cpu$i/CpuX/ if !$cpuEnabled[$i];
$line=~s/Cp$i/CpXX/ if !$cpuEnabled[$i] && length($i)==2;
$line=~s/C$i/CXXX/ if !$cpuEnabled[$i] && length($i)==3;
$line=~s/$i/XXXX/ if !$cpuEnabled[$i] && length($i)==4;
}
}
}
if ($memOpts!~/R/)
{
$line.="Free Buff Cach Inac Slab Map " if $subsys=~/m/;
}
else
{
$line.=" Free Buff Cach Inac Slab Map " if $subsys=~/m/;
}
$line.=" Fragments " if $subsys=~/b/;
$line.=" Alloc Bytes " if $subsys=~/y/ && $slabinfoFlag;
$line.=" InUse Total " if $subsys=~/y/ && $slubinfoFlag;
$line.="KBRead Reads KBWrit Writes " if $subsys=~/d/ && !$ioSizeFlag && $dskOpts!~/i/;
$line.="KBRead Reads Size KBWrit Writes Size " if $subsys=~/d/ && ($ioSizeFlag || $dskOpts=~/i/);
$line.=" KBIn PktIn KBOut PktOut " if $subsys=~/n/ && !$ioSizeFlag && $netOpts!~/i/;
$line.=" KBIn PktIn Size KBOut PktOut Size " if $subsys=~/n/ && ($ioSizeFlag || $netOpts=~/i/);
$line.="Error " if $netOpts=~/e/;
$line.=$tcp2 if $subsys=~/t/;
$line.=" Tcp Udp Raw Frag " if $subsys=~/s/;
$line.="Handle Inodes " if $subsys=~/i/;
$line.=" KBIn PktIn KBOut PktOut Errs " if $subsys=~/x/ && $NumXRails;
$line.=" KBIn PktIn KBOut PktOut Errs " if $subsys=~/x/ && ($NumHCAs || $NumHCAs+$NumXRails==0) && (!$ioSizeFlag && $xOpts!~/i/);
$line.=" KBIn PktIn Size KBOut PktOut Size Errs " if $subsys=~/x/ && ($NumHCAs || $NumHCAs+$NumXRails==0) && ($ioSizeFlag || $xOpts=~/i/);
$line.=" Reads Writes Meta Comm " if $subsys=~/f/;
if ($subsys=~/l/ && $reportMdsFlag)
{
$line.="Gattr+ Sattr+ Sync ";
$line.=($cfsVersion lt '1.6.5') ? 'Reint ' : 'Unlnk ';
}
$line.=" KBRead Reads KBWrit Writes " if $subsys=~/l/ && $reportOstFlag && !$ioSizeFlag;
$line.=" KBRead Reads Size KBWrit Writes Size " if $subsys=~/l/ && $reportOstFlag && $ioSizeFlag;
if ($subsys=~/l/ && $reportCltFlag)
{
$line.=" KBRead Reads KBWrite Writes" if !$ioSizeFlag;
$line.=" KBRead Reads Size KBWrite Writes Size" if $ioSizeFlag;
$line.=" Hits Misses" if $lustOpts=~/R/;
}
for (my $i=0; $i<$impNumMods; $i++) { &{$impPrintBrief[$i]}(2, \$line); }
$line.="\n";
$headersPrinted=1;
if ($showColFlag)
{ printText($line); exit(0); }
}
goto statsSummary if $statsFlag && $statOpts!~/i/i;
# leading space not needed for date/time
$line.=sprintf(' ') if !$miniDateFlag && !$miniTimeFlag;
# First part always the same...
$line.=sprintf("%s ", $datetime) if $miniDateFlag || $miniTimeFlag;
if ($subsys=~/c/)
{
$i=$NumCpus;
$sysTot=$sysP[$i]+$irqP[$i]+$softP[$i]+$stealP[$i];
$cpuTot=$userP[$i]+$niceP[$i]+$sysTot;
$line.=sprintf("%3d %3d %5s %6s ",
$cpuTot, $sysTot, cvt($intrpt/$intSecs,5), cvt($ctxt/$intSecs,6));
}
if ($subsys=~/j/)
{
for (my $i=0; $i<$NumCpus; $i++)
{
$line.=sprintf("%4s ", cvt($intrptTot[$i]/$intSecs,4,0,0));
}
}
if ($subsys=~/m/)
{
if ($memOpts!~/R/)
{
$line.=sprintf("%4s %4s %4s %4s %4s %4s ",
cvt($memFree,4,1,1), cvt($memBuf,4,1,1),
cvt($memCached,4,1,1), cvt($memInact,4,1,1),
cvt($memSlab,4,1,1), cvt($memMap+$memAnon,4,1,1));
}
else
{
$line.=sprintf("%5s %5s %5s %5s %5s %5s ",
cvt($memFreeC/$intSecs,4,1,1), cvt($memBufC/$intSecs,4,1,1),
cvt($memCachedC/$intSecs,4,1,1), cvt($memInactC/$intSecs,4,1,1),
cvt($memSlabC/$intSecs,4,1,1), cvt($memMapC+$memAnonC/$intSecs,4,1,1));
}
}
if ($subsys=~/b/)
{
$line.=sprintf("%s ", base36(@buddyInfoTot));
}
if ($subsys=~/y/)
{
if ($slabinfoFlag)
{
$line.=sprintf("%6s %7s ",
cvt($slabSlabAllTotal,6), cvt($slabSlabAllTotalB,7,0,1));
}
else
{
$line.=sprintf("%6s %7s ",
cvt($slabNumObjTot,7), cvt($slabTotalTot,7,0,1));
}
}
if ($subsys=~/d/)
{
if (!$ioSizeFlag && $dskOpts!~/i/)
{
$line.=sprintf("%6s %6s %6s %6s ",
cvt($dskReadKBTot/$intSecs,6,0,1), cvt($dskReadTot/$intSecs,6),
cvt($dskWriteKBTot/$intSecs,6,0,1), cvt($dskWriteTot/$intSecs,6));
}
else
{
$dskReadSizeTot= ($dskReadTot) ? $dskReadKBTot/$dskReadTot : 0;
$dskWriteSizeTot=($dskWriteTot) ? $dskWriteKBTot/$dskWriteTot : 0;
$line.=sprintf("%6s %6s %4s %6s %6s %4s ",
cvt($dskReadKBTot/$intSecs,6,0,1), cvt($dskReadTot/$intSecs,6), cvt($dskReadSizeTot, 4),
cvt($dskWriteKBTot/$intSecs,6,0,1), cvt($dskWriteTot/$intSecs,6), cvt($dskWriteSizeTot, 4));
}
}
# Network always the same
my $netErrors=$netRxErrsTot+$netTxErrsTot;
if ($subsys=~/n/)
{
if (!$ioSizeFlag && $netOpts!~/i/)
{
$line.=sprintf("%6s %6s %6s %6s ",
cvt($netRxKBTot/$intSecs,6,0,1), cvt($netRxPktTot/$intSecs,6),
cvt($netTxKBTot/$intSecs,6,0,1), cvt($netTxPktTot/$intSecs,6));
}
else
{
$netRxSizeTot=($netRxPktTot) ? $netRxKBTot*1024/$netRxPktTot : 0;
$netTxSizeTot=($netTxPktTot) ? $netTxKBTot*1024/$netTxPktTot : 0;
$line.=sprintf("%6s %6s %4s %6s %6s %4s ",
cvt($netRxKBTot/$intSecs,6,0,1), cvt($netRxPktTot/$intSecs,6), cvt($netRxSizeTot,4,0,1),
cvt($netTxKBTot/$intSecs,6,0,1), cvt($netTxPktTot/$intSecs,6), cvt($netTxSizeTot,4,0,1));
}
# if --netops E and no errors, don't print ANYTHING!!!
$line.=sprintf("%5s ", cvt($netErrors/$intSecs,5)) if $netOpts=~/e/;
}
# TCP Stack
if ($subsys=~/t/)
{
$line.=sprintf("%4s ", cvt($ipErrors, 4)) if $tcpFilt=~/i/;
$line.=sprintf("%4s ", cvt($tcpErrors, 4)) if $tcpFilt=~/t/;
$line.=sprintf("%4s ", cvt($udpErrors, 4)) if $tcpFilt=~/u/;
$line.=sprintf("%4s ", cvt($icmpErrors, 4)) if $tcpFilt=~/c/;
$line.=sprintf("%4s ", cvt($tcpExErrors,4)) if $tcpFilt=~/T/;
}
if ($subsys=~/s/)
{
$line.=sprintf("%4s %4s %4s %4s ",
cvt($sockUsed,4), cvt($sockUdp,4), cvt($sockRaw,4), cvt($sockFrag,4));
}
if ($subsys=~/i/)
{
$line.=sprintf("%6s %6s ", cvt($filesAlloc, 6), cvt($inodeUsed, 6));
}
# and so is elan
if ($subsys=~/x/)
{
if ($NumXRails)
{
$elanErrors=$elanSendFailTot+$elanNeterrAtomicTot+$elanNeterrDmaTot;
$line.=sprintf("%7s %6s %7s %6s %4s ",
cvt($elanRxKBTot/$intSecs,7,0,1), cvt($elanRxTot/$intSecs,6),
cvt($elanTxKBTot/$intSecs,7,0,1), cvt($elanTxTot/$intSecs,6),
cvt($elanErrors/$intSecs,4));
}
if ($NumHCAs || $NumXRails+$NumHCAs==0)
{
if (!$ioSizeFlag && $xOpts!~/i/)
{
$line.=sprintf("%7s %6s %7s %6s %4s ",
cvt($ibRxKBTot/$intSecs,7,0,1), cvt($ibRxTot/$intSecs,6),
cvt($ibTxKBTot/$intSecs,7,0,1), cvt($ibTxTot/$intSecs,6),
cvt($ibErrorsTotTot,4));
}
else
{
$line.=sprintf("%7s %6s %4s %7s %6s %4s %4s ",
cvt($ibRxKBTot/$intSecs,7,0,1), cvt($ibRxTot/$intSecs,6),
$ibRxTot ? cvt($ibRxKBTot*1024/$ibRxTot,4,0,1) : 0,
cvt($ibTxKBTot/$intSecs,7,0,1), cvt($ibTxTot/$intSecs,6),
$ibTxTot ? cvt($ibTxKBTot*1024/$ibTxTot,4,0,1) : 0,
cvt($ibErrorsTotTot,4));
}
}
}
if ($subsys=~/f/)
{
$line.=sprintf("%6s %6s %4s %4s ",
cvt($nfsReadsTot/$intSecs,6), cvt($nfsWritesTot/$intSecs,6),
cvt($nfsMetaTot/$intSecs), cvt($nfsCommitTot/$intSecs));
}
# MDS
if ($subsys=~/l/ && $reportMdsFlag)
{
my $setattrPlus=$lustreMdsReintSetattr+$lustreMdsSetxattr;
my $getattrPlus=$lustreMdsGetattr+$lustreMdsGetattrLock+$lustreMdsGetxattr;
my $variableParam=($cfsVersion lt '1.6.5') ? $lustreMdsReint : $lustreMdsReintUnlink;
$line.=sprintf("%6s %6s %6s %6s ",
cvt($getattrPlus/$intSecs,6), cvt($setattrPlus/$intSecs,6),
cvt($lustreMdsSync/$intSecs,6), cvt($variableParam/$intSecs,6));
}
# OST
if ($subsys=~/l/ && $reportOstFlag)
{
if (!$ioSizeFlag)
{
$line.=sprintf("%7s %6s %7s %6s ",
cvt($lustreReadKBytesTot/$intSecs,7,0,1), cvt($lustreReadOpsTot/$intSecs,6),
cvt($lustreWriteKBytesTot/$intSecs,7,0,1), cvt($lustreWriteOpsTot/$intSecs,6));
}
else
{
$line.=sprintf("%7s %6s %4s %7s %6s %4s ",
cvt($lustreReadKBytesTot/$intSecs,7,0,1), cvt($lustreReadOpsTot/$intSecs,6),
$lustreReadOpsTot ? cvt($lustreReadKBytesTot/$lustreReadOpsTot,4,0,1) : 0,
cvt($lustreWriteKBytesTot/$intSecs,7,0,1), cvt($lustreWriteOpsTot/$intSecs,6),
$lustreWriteOpsTot ? cvt($lustreWriteKBytesTot/$lustreWriteOpsTot,4,0,1) : 0);
}
}
#Lustre Client
if ($subsys=~/l/ && $reportCltFlag)
{
if (!$ioSizeFlag)
{
$line.=sprintf("%7s %6s %7s %6s",
cvt($lustreCltReadKBTot/$intSecs,7,0,1), cvt($lustreCltReadTot/$intSecs),
cvt($lustreCltWriteKBTot/$intSecs,7,0,1), cvt($lustreCltWriteTot/$intSecs,6));
}
else
{
$line.=sprintf("%7s %6s %4s %7s %6s %4s",
cvt($lustreCltReadKBTot/$intSecs,7,0,1), cvt($lustreCltReadTot/$intSecs),
$lustreCltReadTot ? cvt($lustreCltReadKBTot/$lustreCltReadTot,4,0,1) : 0,
cvt($lustreCltWriteKBTot/$intSecs,7,0,1), cvt($lustreCltWriteTot/$intSecs,6),
$lustreCltWriteTot ? cvt($lustreCltWriteKBTot/$lustreCltWriteTot,4,0,1) : 0);
}
# Add in cache hits/misses if --lustopts R
$line.=sprintf(" %6d %6d", $lustreCltRAHitsTot, $lustreCltRAMissesTot) if $lustOpts=~/R/;
}
for (my $i=0; $i<$impNumMods; $i++) { &{$impPrintBrief[$i]}(3, \$line); }
$line.="\n";
# S p e c i a l ' h o t ' K e y P r o c e s s i n g
# First time through when an attached terminal
if ($termFlag && !defined($mini1select))
{
$mini1select=new IO::Select(STDIN);
resetBriefCounters();
`stty -echo` if !$PcFlag && !$backFlag && $termFlag && $playback eq '';
}
# See if user entered a command. If not, @ready will never be
# non-zero so the 'if' below will never fire. Also, if we haven't
# done one interval, ignore becuase $miniInstances will be 0
@ready=$mini1select->can_read(0) if $termFlag;
if (scalar(@ready))
{
$command=;
if ($miniInstances)
{
$resetType='T';
$resetType=$command if $command=~/a|t|z/i;
printBriefCounters($resetType);
resetBriefCounters() if $resetType=~/Z/i;
}
}
# come here from collectl's ONLY goto statement!
statsSummary:
# Minor subtlety - we want to print the totals as soon as the hot-key
# is entered and so we print the sub-total so far which DOESN'T
# include this latest line! Then we count the data.
countBriefCounters();
$miniInstances++;
# The only time we don't print the line is if it doesn't contain any data, which should only happen
# when data was imported at a different interval and played back with -s-all, OR we're only
# doing network error reporting and this interval is clean. In that cast reset '$totalCounter'
# so header printing works correctly.
$empty=0;
$empty=1 if $import ne '' && $subsys eq '' && length($line) == length($datetime)+1;
printText($line) if !$empty && ($netOpts!~/E/ || $netErrors);
$totalCounter-- if $netOpts=~/E/ && !$netErrors
}
sub resetBriefCounters
{
# talk about a mouthful!
$miniStart=0;
$miniInstances=0;
$cpuTOT=$sysPTOT=$intrptTOT=$ctxtTOT=0;
$memFreeTOT=$memBufTOT=$memCachedTOT=$memInactTOT=$memSlabTOT=$memMapTOT=0;
$memFreeCTOT=$memBufCTOT=$memCachedCTOT=$memInactCTOT=$memSlabCTOT=$memMapCTOT=0;
$slabSlabAllTotalTOT=$slabSlabAllTotalBTOT=0;
$dskReadKBTOT=$dskReadTOT=$dskWriteKBTOT=$dskWriteTOT=0;
$netRxKBTOT=$netRxPktTOT=$netTxKBTOT=$netTxPktTOT=$netErrTOT=0;
$tcpIpErrTOT=$tcpIcmpErrTOT=$tcpTcpErrTOT=$tcpUdpErrTOT=$tcpTcpExErrTOT=0;
$sockUsedTOT=$sockUdpTOT=$sockRawTOT=$sockFragTOT=0;
$filesAllocTOT=$inodeUsedTOT=0;
$elanRxKBTOT=$elanRxTOT=$elanTxKBTOT=$elanTxTOT=$elanErrorsTOT=0;
$ibRxKBTOT=$ibRxTOT=$ibTxKBTOT=$ibTxTOT=$ibErrorsTOT=0;
$nfsReadsTOT=$nfsWritesTOT=$nfsMetaTOT=$nfsCommitTOT=0;
$lustreMdsGetattrPlusTOT=$lustreMdsSetattrPlusTOT=$lustreMdsSyncTOT=0;
$lustreMdsReintTOT=$lustreMdsReintUnlinkTOT=0;
$lustreReadKBytesTOT=$lustreReadOpsTOT=$lustreWriteKBytesTOT=$lustreWriteOpsTOT=0;
$lustreCltReadTOT=$lustreCltReadKBTOT=$lustreCltWriteTOT=$lustreCltWriteKBTOT=0;
$lustreCltRAHitsTOT=$lustreCltRAMissesTOT=0;
for (my $i=0; $i<$numBrwBuckets; $i++)
{ $lustreBufReadTOT[$i]=$lustreBufWriteTOT[$i]=0; }
for (my $i=0; $i<$NumCpus; $i++)
{ $intrptTOT[$i]=0; }
for (my $i=0; $i<11; $i++)
{ $buddyInfoTOT[$i]=0; }
for (my $i=0; $i<$impNumMods; $i++) { &{$impPrintBrief[$i]}(4); }
}
sub countBriefCounters
{
my $i=$NumCpus;
$cpuTOT+= $userP[$i]+$niceP[$i]+$sysP[$i];
$sysPTOT+= $sysP[$i];
$intrptTOT+=$intrpt;
$ctxtTOT+= $ctxt;
for ($i=0; $i<$NumCpus; $i++)
{ $intrptTOT[$i]+=$intrptTot[$i]; }
# the default, it so add up the amount of actual memory used
# could have reused the TOT counter names, but let's not
if ($memOpts!~/R/)
{
$memFreeTOT+= $memFree;
$memBufTOT+= $memBuf;
$memCachedTOT+=$memCached;
$memInactTOT+= $memInact;
$memSlabTOT+= $memSlab;
$memMapTOT+= $memMap+$memAnon;
}
else # in this case we add up the changes
{
$memFreeCTOT+= $memFreeC;
$memBufCTOT+= $memBufC;
$memCachedCTOT+=$memCachedC;
$memInactCTOT+= $memInactC;
$memSlabCTOT+= $memSlabC;
$memMapCTOT+= $memMapC+$memAnonC;
}
$slabSlabAllTotalTOT+= $slabSlabAllTotal;
$slabSlabAllTotalBTOT+=$slabSlabAllTotalB;
$dskReadKBTOT+= $dskReadKBTot;
$dskReadTOT+= $dskReadTot;
$dskWriteKBTOT+= $dskWriteKBTot;
$dskWriteTOT+= $dskWriteTot;
$netRxKBTOT+= $netRxKBTot;
$netRxPktTOT+= $netRxPktTot;
$netTxKBTOT+= $netTxKBTot;
$netTxPktTOT+= $netTxPktTot;
$netErrTOT+= $netRxErrsTot+$netTxErrsTot;
$tcpIpErrTOT+= $ipErrors;
$tcpIcmpErrTOT+= $icmpErrors;
$tcpTcpErrTOT+= $tcpErrors;
$tcpUdpErrTOT+= $udpErrors;
$tcpTcpExErrTOT+=$tcpExErrors;
$sockUsedTOT+= $sockUsed;
$sockUdpTOT+= $sockUdp;
$sockRawTOT+= $sockRaw;
$sockFragTOT+= $sockFrag;
$filesAllocTOT+= $filesAlloc;
$inodeUsedTOT+= $inodeUsed;
$elanRxKBTOT+= $elanRxKBTot;
$elanRxTOT+= $elanRxTot;
$elanTxKBTOT+= $elanTxKBTot;
$elanTxTOT+= $elanTxTot;
$elanErrorsTOT+= $elanErrors;
$ibRxKBTOT+= $ibRxKBTot;
$ibRxTOT+= $ibRxTot;
$ibTxKBTOT+= $ibTxKBTot;
$ibTxTOT+= $ibTxTot;
$ibErrorsTOT+= $ibErrorsTotTot;
$nfsReadsTOT+= $nfsReadsTot;
$nfsWritesTOT+= $nfsWritesTot;
$nfsMetaTOT+= $nfsMetaTot;
$nfsCommitTOT+= $nfsCommitTot;
if ($NumMds)
{
# Although some apply to versions < 1.6.5, easier to just count everything
$lustreMdsGetattrPlusTOT+=$lustreMdsGetattr+$lustreMdsGetattrLock+$lustreMdsGetxattr;
$lustreMdsSetattrPlusTOT+=$lustreMdsReintSetattr+$lustreMdsSetxattr;
$lustreMdsSyncTOT+= $lustreMdsSync;
$lustreMdsReintTOT+= $lustreMdsReint;
$lustreMdsReintUnlinkTOT+=$lustreMdsReintUnlink;
}
if ($NumOst)
{
$lustreReadKBytesTOT+= $lustreReadKBytesTot;
$lustreReadOpsTOT+= $lustreReadOpsTot;
$lustreWriteKBytesTOT+=$lustreWriteKBytesTot;
$lustreWriteOpsTOT+= $lustreWriteOpsTot;
}
if ($reportCltFlag)
{
$lustreCltReadTOT+= $lustreCltReadTot;
$lustreCltReadKBTOT+= $lustreCltReadKBTot;
$lustreCltWriteTOT+= $lustreCltWriteTot;
$lustreCltWriteKBTOT+=$lustreCltWriteKBTot;
$lustreCltRAHitsTOT+= $lustreCltRAHitsTot;
$lustreCltRAMissesTOT+=$lustreCltRAMissesTot;
}
if ($NumBud)
{
for ($i=0; $i<11; $i++)
{
$buddyInfoTOT[$i]+=$buddyInfoTot[$i];
}
}
for (my $i=0; $i<$impNumMods; $i++) { &{$impPrintBrief[$i]}(5); }
}
sub printBriefCounters
{
my $type=shift;
my $i;
# For things that totals don't make sense, like CPUs or sockets, just do averags all the time
# by using the number of instances
my $mi=$miniInstances;
my $totSecs=$interval; # makes calculation of total easy
if ($type=~/a/i)
{
# Totals are NOT normalized so for averages we need to divide by total seconds.
$totSecs=($playback eq '') ? $seconds-$miniStart+$interval : $elapsedSecs;
$datetime=' ' x length($datetime) if $statOpts!~/s/i; # when not in summary mode, include date/time stamps
}
chomp $type;
printf "%s", $datetime if $miniDateFlag || $miniTimeFlag;
printf "%s", uc($type);
printf "%3d %3d %5s %6s ",
$cpuTOT/$mi, $sysPTOT/$mi, cvt($intrptTOT/$totSecs,5), cvt($ctxtTOT/$totSecs,6)
if $subsys=~/c/;
if ($subsys=~/j/)
{
for (my $i=0; $i<$NumCpus; $i++)
{
printf "%4s ", cvt($intrptTOT[$i]/$totSecs,4,0,0);
}
}
if ($subsys=~/m/)
{
if ($memOpts!~/R/)
{
printf "%4s %4s %4s %4s %4s %4s ",
cvt($memFreeTOT/$mi,4,1,1), cvt($memBufTOT/$mi,4,1,1), cvt($memCachedTOT/$mi,4,1,1),
cvt($memInactTOT/$mi,4,1,1), cvt($memSlabTOT/$mi,4,1,1), cvt($memMapTOT/$mi,4,1,1);
}
else
{
printf "%5s %5s %5s %5s %5s %5s ",
cvt($memFreeCTOT/$mi,4,1,1), cvt($memBufCTOT/$mi,4,1,1), cvt($memCachedCTOT/$mi,4,1,1),
cvt($memInactCTOT/$mi,4,1,1), cvt($memSlabCTOT/$mi,4,1,1), cvt($memMapCTOT/$mi,4,1,1);
}
}
# Need to average each field before converting
if ($subsys=~/b/)
{
for ($i=0; $i<11; $i++)
{ $buddyInfoAVG[$i]=$buddyInfoTOT[$i]/$mi; }
printf "%s ", base36(@buddyInfoAVG);
}
# Will probably never be used again
printf "%6s %7s ",
cvt($slabSlabAllTotalTOT/$mi,6,0,1), cvt($slabSlabAllTotalBTOT/$mi,7,0,1)
if $subsys=~/y/;
if ($subsys=~/d/)
{
if (!$ioSizeFlag && $dskOpts!~/i/)
{
printf "%6s %6s %6s %6s ",
cvt($dskReadKBTOT/$totSecs,6,0,1), cvt($dskReadTOT/$totSecs,6),
cvt($dskWriteKBTOT/$totSecs,6,0,1), cvt($dskWriteTOT/$totSecs,6);
}
else
{
printf "%6s %6s %4s %6s %6s %4s ",
cvt($dskReadKBTOT/$totSecs,6,0,1), cvt($dskReadTOT/$totSecs,6),
$dskReadTOT ? cvt($dskReadKBTOT/$dskReadTOT,4,0,1) : 0,
cvt($dskWriteKBTOT/$totSecs,6,0,1), cvt($dskWriteTOT/$totSecs,6),
$dskWriteTOT ? cvt($dskWriteKBTOT/$dskWriteTOT,4,0,1) : 0;
}
}
if ($subsys=~/n/)
{
if (!$ioSizeFlag && $netOpts!~/i/)
{
printf "%6s %6s %6s %6s ",
cvt($netRxKBTOT/$totSecs,6,0,1), cvt($netRxPktTOT/$totSecs,6),
cvt($netTxKBTOT/$totSecs,6,0,1), cvt($netTxPktTOT/$totSecs,6);
}
else
{
printf "%6s %6s %4s %6s %6s %4s ",
cvt($netRxKBTOT/$totSecs,6,0,1), cvt($netRxPktTOT/$totSecs,6),
$netRxPktTOT ? cvt($netRxKBTOT*1024/$netRxPktTOT,4,0,1) : 0,
cvt($netTxKBTOT/$totSecs,6,0,1), cvt($netTxPktTOT/$totSecs,6),
$netTxPktTOT ? cvt($netTxKBTOT*1024/$netTxPktTOT,4,0,1) : 0;
}
printf "%5s ", cvt($netErrTOT/$totSecs,5) if $netOpts=~/e/;
}
if ($subsys=~/t/)
{
printf "%4s ", cvt($tcpIpErrTOT/$totSecs,4) if $tcpFilt=~/i/;
printf "%4s ", cvt($tcpTcpErrTOT/$totSecs,4) if $tcpFilt=~/t/;
printf "%4s ", cvt($tcpUdpErrTOT/$totSecs,4) if $tcpFilt=~/u/;
printf "%4s ", cvt($tcpIcmpErrTOT/$totSecs,4) if $tcpFilt=~/c/;
printf "%4s ", cvt($tcpTcpExErrTOT/$totSecs,4) if $tcpFilt=~/T/;
}
printf "%4d %4d %4d %4d ",
cvt(int($sockUsedTOT/$mi),6), cvt(int($sockUdpTOT/$mi),6),
cvt(int($sockRawTOT/$mi),6), cvt(int($sockFragTOT/$mi),6)
if $subsys=~/s/;
printf "%6s %6s ", cvt($filesAllocTOT/$mi, 6), cvt($inodeUsedTOT/$mi, 6)
if $subsys=~/i/;
printf "%7s %6s %7s %6s %6s ",
cvt($elanRxKBTOT/$totSecs,6,0,1), cvt($elanRxTOT/$totSecs,6),
cvt($elanTxKBTOT/$totSecs,6,0,1), cvt($elanTxTOT/$totSecs,6),
cvt($elanErrorsTOT/$totSecs,6)
if $subsys=~/x/ && $NumXRails;
if ($subsys=~/x/ && $NumHCAs)
{
if (!$ioSizeFlag && $xOpts!~/i/)
{
printf "%7s %6s %7s %6s %4s ",
cvt($ibRxKBTOT/$totSecs,7,0,1), cvt($ibRxTOT/$totSecs,6),
cvt($ibTxKBTOT/$totSecs,7,0,1), cvt($ibTxTOT/$totSecs,6),
cvt($ibErrorsTOT,4);
}
else
{
printf "%7s %6s %4s %7s %6s %4s %4s ",
cvt($ibRxKBTOT/$totSecs,7,0,1), cvt($ibRxTOT/$totSecs,6),
$ibRxTOT ? cvt($ibRxKBTOT*1024/ibRxTOT,4,0,1) : 0,
cvt($ibTxKBTOT/$totSecs,7,0,1), cvt($ibTxTOT/$totSecs,6),
$ibTxTOT ? cvt($ibTxKBTOT*1024/ibTxTOT,4,0,1) : 0,
cvt($ibErrorsTOT,4);
}
}
printf "%6s %6s %4s %4s ",
cvt($nfsReadsTOT/$totSecs,6), cvt($nfsWritesTOT/$totSecs,6),
cvt($nfsMetaTOT/$totSecs), cvt($nfsCommitTOT/$totSecs)
if $subsys=~/f/;
if ($subsys=~/l/ && $reportMdsFlag)
{
my $variableParam=($cfsVersion lt '1.6.5') ? $lustreMdsReintTOT : $lustreMdsReintUnlinkTOT;
printf "%6s %6s %6s %6s ",
cvt($lustreMdsGetattrPlusTOT/$totSecs,6), cvt($lustreMdsSetattrPlusTOT/$totSecs,6),
cvt($lustreMdsSyncTOT/$totSecs,6), cvt($variableParam/$totSecs,6);
}
if ($subsys=~/l/ && $reportOstFlag)
{
if (!$ioSizeFlag)
{
printf "%7s %6s %7s %6s ",
cvt($lustreReadKBytesTOT/$totSecs,7,0,1), cvt($lustreReadOpsTOT/$totSecs,6),
cvt($lustreWriteKBytesTOT/$totSecs,7,0,1), cvt($lustreWriteOpsTOT/$totSecs,6);
}
else
{
printf "%7s %6s %4s %7s %6s %4s ",
cvt($lustreReadKBytesTOT/$totSecs,7,0,1), cvt($lustreReadOpsTOT/$totSecs,6),
$lustreReadOpsTOT ? cvt($lustreReadKBytesTOT/$lustreReadOpsTOT,4,0,1) : 0,
cvt($lustreWriteKBytesTOT/$totSecs,7,0,1), cvt($lustreWriteOpsTOT/$totSecs,6),
$lustreWriteOpsTOT ? cvt($lustreWriteKBytesTOT/$lustreWriteOpsTOT,4,0,1) : 0;
}
}
if ($subsys=~/l/ && $reportCltFlag)
{
if (!$ioSizeFlag)
{
printf "%7s %6s %7s %6s",
cvt($lustreCltReadKBTOT/$totSecs,7,0,1), cvt($lustreCltReadTOT/$totSecs,6),
cvt($lustreCltWriteKBTOT/$totSecs,7,0,1), cvt($lustreCltWriteTOT/$totSecs,6);
}
else
{
printf "%7s %6s %4s %7s %6s %4s",
cvt($lustreCltReadKBTOT/$totSecs,7,0,1), cvt($lustreCltReadTOT/$totSecs,6),
$lustreCltReadTOT ? cvt($lustreCltReadKBTOT/$lustreCltReadTOT,4,0,1) : 0,
cvt($lustreCltWriteKBTOT/$totSecs,7,0,1), cvt($lustreCltWriteTOT/$totSecs,6),
$lustreCltWriteTOT ? cvt($lustreCltWriteKBTOT/$lustreCltWriteTOT,4,0,1) : 0;
}
printf " %6s %6s",
cvt($lustreCltRAHitsTOT/$totSecs,6),cvt($lustreCltRAMissesTOT/$totSecs,6)
if $lustOpts=~/R/;
}
for (my $i=0; $i<$impNumMods; $i++) { &{$impPrintBrief[$i]}(6); }
print "\n";
}
sub base36
{
my @buddies=@_;
my $frags;
for (my $i=0; $i=1000)
{
# 1000->the res => 30->36
$map=int(log($num)/log(10))-3;
$map=8 if $map>8;
$frag=substr('stuvwxyz', $map, 1);
}
elsif ($num>=100)
{
# 100->999 => 20->29
$map=int($num)/100-1;
$frag=substr('jklmnopqr', $map, 1);
}
elsif ($num>=10)
{
# 10->99 => 10->19
$map=int($num)/10-1;
$frag=substr('abcdefghi', $map, 1);
}
else
{
# 0->9 => 0->9
$frag=int($num);
}
$frags.=$frag;
}
return($frags);
}
####################################################
# T a s k P r o c e s s i n g S u p p o r t
####################################################
sub nextAvailProcIndex
{
my $next;
if (scalar(@procIndexesFree)>0)
{ $next=pop @procIndexesFree; }
else
{ $next=$procNextIndex++; }
printf "### Index allocated: $next NextIndex: $procNextIndex IndexesFree: %d\n",
scalar(@procIndexesFree) if $debug & 256;
return($next);
}
# If we're not processing by pid-only, the processes we're reporting on come
# and go. Therefore right before we print we need to see if a process we
# were reporting on disappeared by noticing its pid went away and therefore
# need to remove it from the $procIndexes{} hash. Is there a better/more
# efficient way to do this? If so, fix 'cleanStalePids()' too.
sub cleanStaleTasks
{
my ($removeFlag, %indexesTemp, $pid);
if ($debug & 512)
{
print "### CleanStaleTasks()\n";
foreach $pid (sort keys %procSeen)
{ print "### PIDPROC: $pid\n"; }
}
# make a list of only those pids we've seen during last cycle
$removeFlag=0;
foreach $pid (sort keys %procIndexes)
{
if (defined($procSeen{$pid}))
{
$indexesTemp{$pid}=$procIndexes{$pid};
print "### indexesTemp[$pid] set to $indexesTemp{$pid}\n"
if $debug & 256;
}
else
{
push @procIndexesFree, $procIndexes{$pid};
$removeFlag=1;
print "### added $pid with index of $procIndexes{$pid} to free list\n"
if $debug & 256;
}
}
# only need to do a swap if we need to remove a pid.
if ($removeFlag)
{
undef %procIndexes;
%procIndexes=%indexesTemp;
if ($debug & 512)
{
print "### Indexes Swapped! NEW procIndexes{}\n";
foreach $key (sort keys %procIndexes)
{ print "procIndexes{$key}=$procIndexes{$key}\n"; }
}
}
undef %procSeen;
}
# This output only goes to the .prc file
sub printPlotProc
{
my $date=shift;
my $time=shift;
my ($procHeaders, $procPlot, $pid, $i);
$procHeaders='';
if (!$headersPrintedProc)
{
$procHeaders=$commonHeader if $logToFileFlag;
$procHeaders.=(!$utcFlag) ? "#Date${SEP}Time" : '#UTC';;
$procHeaders.="${SEP}PID${SEP}User${SEP}PR${SEP}PPID${SEP}THRD${SEP}S${SEP}VmSize${SEP}";
$procHeaders.="VmLck${SEP}VmRSS${SEP}VmData${SEP}VmStk${SEP}VmExe${SEP}VmLib${SEP}";
$procHeaders.="CPU${SEP}SysT${SEP}UsrT${SEP}PCT${SEP}AccumT${SEP}";
$procHeaders.="RKB${SEP}WKB${SEP}RKBC${SEP}WKBC${SEP}RSYS${SEP}WSYS${SEP}CNCL${SEP}";
$procHeaders.="MajF${SEP}MinF${SEP}Command\n";
$headersPrintedProc=1;
}
$procPlot=$procHeaders;
foreach $pid (sort keys %procIndexes)
{
$i=$procIndexes{$pid};
next if (!defined($procSTimeTot[$i]));
next if $procState ne '' && $procState[$i]!~/[$procState]/;
# Handle -oF
if ($procOpts=~/f/)
{
$majFlt=$procMajFltTot[$i];
$minFlt=$procMinFltTot[$i];
}
else
{
$majFlt=$procMajFlt[$i]/$interval2Secs;
$minFlt=$procMinFlt[$i]/$interval2Secs;
}
my $datetime=(!$utcFlag) ? "$date$SEP$time": time;
$datetime.=".$usecs" if $options=~/m/;
# Username comes from translation hash OR we just print the UID
$procPlot.=sprintf("%s${SEP}%d${SEP}%s${SEP}%s${SEP}%s${SEP}%d${SEP}%s${SEP}%s${SEP}%s${SEP}%s${SEP}%s${SEP}%s${SEP}%s${SEP}%s${SEP}%s${SEP}%s${SEP}%s${SEP}%d${SEP}%s${SEP}%d${SEP}%d${SEP}%d${SEP}%d${SEP}%d${SEP}%d${SEP}%d${SEP}%s${SEP}%s${SEP}%s",
$datetime, $procPid[$i], $procUser[$i], $procPri[$i],
$procPpid[$i], $procThread[%i], $procState[$i],
defined($procVmSize[$i]) ? $procVmSize[$i] : 0,
defined($procVmLck[$i]) ? $procVmLck[$i] : 0,
defined($procVmRSS[$i]) ? $procVmRSS[$i] : 0,
defined($procVmData[$i]) ? $procVmData[$i] : 0,
defined($procVmStk[$i]) ? $procVmStk[$i] : 0,
defined($procVmExe[$i]) ? $procVmExe[$i] : 0,
defined($procVmLib[$i]) ? $procVmLib[$i] : 0,
$procCPU[$i],
cvtT1($procSTime[$i],1), cvtT1($procUTime[$i],1),
($procSTime[$i]+$procUTime[$i])/$interval2SecsReal,
cvtT2($procSTimeTot[$i]+$procUTimeTot[$i],1),
defined($procRKB[$i]) ? $procRKB[$i]/$interval2Secs : 0,
defined($procWKB[$i]) ? $procWKB[$i]/$interval2Secs : 0,
defined($procRKBC[$i]) ? $procRKBC[$i]/$interval2Secs : 0,
defined($procWKBC[$i]) ? $procWKBC[$i]/$interval2Secs : 0,
defined($procRSys[$i]) ? $procRSys[$i]/$interval2Secs : 0,
defined($procWSys[$i]) ? $procWSys[$i]/$interval2Secs : 0,
defined($procCKB[$i]) ? $procCKB[$i]/$interval2Secs : 0,
cvt($majFlt), cvt($minFlt),
defined($procCmd[$i]) ? $procCmd[$i] : $procName[$i]);
# This is a little messy (sorry about that). The way writeData works is that
# on writeData(0) calls, it builds up a string in $oneline which can be appended
# to the current string (for displaying multiple subsystems in plot format on
# the terminal and the final call writes it out. In order for all the paths
# to work with sockets, etc we need to do it this way. And since writeData takes
# care of \n be sure to leave OFF each line being written.
$oneline='';
writeData(0, '', \$procPlot, PRC, $ZPRC, 'proc', \$oneline);
if (!$logToFileFlag || ($sockFlag && $export eq ''))
{
last if writeData(1, '', undef, $LOG, undef, undef, \$oneline)==0;
}
$procPlot='';
}
}
sub procAnalyze
{
my $seconds=shift;
my $usecs= shift;
my ($vmSize, $vmLck, $vmRSS, $vmData, $vmStk, $vmLib, $vmExe);
my ($rkb, $wkb, $rkbc, $wkbc, $rsys, $wsys, $cncl, $threads);
# Would have been nice to use $interval2Counter, but that only increments
# during terminal output.
$procAnalCounter++;
# loops through all processes for this interval and copy data to simpler variables
foreach my $pid (keys %procIndexes)
{
# Global which indicates at least 1 piece of process data recorded.
# we also need to save pids so we'll know what to print to file
$procAnalyzed=1;
$analyzed{$pid}=1;
my $i=$procIndexes{$pid};
my $user=$procUser[$i];
my $ppid=$procPpid[$i];
my $threads=$procTCount[$i];
my $cpu=$procCPU[$i];
my $sysT=$procSTime[$i];
my $usrT=$procUTime[$i];
my $accum=cvtT2($procSTimeTot[$i]+$procUTimeTot[$i]);
my $majF=$procMajFlt[$i];
my $minF=$procMinFlt[$i];
my $command=(defined($procCmd[$i])) ? $procCmd[$i] : $procName[$i];
$accum=~s/^\s*//g;
$command=~s/\s+$//g;
if (defined($procVmSize[$i]))
{
$vmSize=$procVmSize[$i];
$vmLck=$procVmLck[$i];
$vmRSS=$procVmRSS[$i];
$vmData=$procVmData[$i];
$vmStk=$procVmStk[$i];
$vmLib=$procVmLib[$i];
$vmExe=$procVmExe[$i];
}
else
{
$vmSize=$vmLck=$vmRSS=$vmData=$vmStk=$vmLib=$vmExe=0;
}
if ($processIOFlag)
{
$rkb=$procRKB[$i]; $wkb=$procWKB[$i]; $rkbc=$procRKBC[$i]; $wkbc=$procWKBC[$i];
$rsys=$procRSys[$i]; $wsys=$procWSys[$i]; $cncl=$procCKB[$i];
}
# Here's what's going on. We're identifying a unique command by its pid and
# name. That way if pids are reused the probability of the same pid showing
# up for the same command are slim. BUT, when processing multiple logs for
# the same day it CAN happen, so we're adding a filename discriminator as well.
my $unique="$fileRoot:$pid:$command";
if (!defined($summary[$pid]))
{
$summary[$pid]={
date=>$date, timefrom=>$seconds, threadsMin=>$threads, threadsMax=>$threads,
pid=>$pid, user=>$user, ppid=>$ppid, vmExe=>$vmExe,
vmSizeMin=>$vmSize, vmSizeMax=>$vmSize, vmLckMin=>$vmLck, vmLckMax=>$vmLck,
vmRSSMin=>$vmRSS, vmRSSMax=>$vmRSS, vmDataMin=>$vmData, vmDataMax=>$vmData,
vmStkMin=>$vmStk, vmStkMax=>$vmStk, vmLibMin=>$vmLib, vmLibMax=>$vmLib,
sysT=>0, usrT=>0, majF=>0, minF=>0, RKB=>0, WKB=>0, RKBC=>0,
WKBC=>0, RSYS=>0, WSYS=>0, CNCL=>0, command=>$command,
timethru=>$seconds, accumT=>0
}
}
next if $procAnalCounter==1;
# U p d a t e S u m m a r y
# note - we also initialized timethru above in case only a single sample
$summary[$pid]->{timethru}=$seconds;
# thread counts not necessarily included in raw file
$summary[$pid]->{threadsMin}=$threads if defined($threads) && $threads<$summary[$pid]->{threadsMin};
$summary[$pid]->{threadsMax}=$threads if defined($threads) && $threads>$summary[$pid]->{threadsMax};
$summary[$pid]->{vmSizeMin}=$vmSize if $vmSize<$summary[$pid]->{vmSizeMin};
$summary[$pid]->{vmSizeMax}=$vmSize if $vmSize>$summary[$pid]->{vmSizeMax};
$summary[$pid]->{vmLckMin}=$vmLck if $vmLck< $summary[$pid]->{vmLckMin};
$summary[$pid]->{vmLckMax}=$vmLck if $vmLck> $summary[$pid]->{vmLckMax};
$summary[$pid]->{vmRSSMin}=$vmRSS if $vmRSS< $summary[$pid]->{vmRSSMin};
$summary[$pid]->{vmRSSMax}=$vmRSS if $vmRSS> $summary[$pid]->{vmRSSMax};
$summary[$pid]->{vmDataMin}=$vmData if $vmData<$summary[$pid]->{vmDataMin};
$summary[$pid]->{vmDataMax}=$vmData if $vmData>$summary[$pid]->{vmDataMax};
$summary[$pid]->{vmStkMin}=$vmStk if $vmStk< $summary[$pid]->{vmStkMin};
$summary[$pid]->{vmStkMax}=$vmStk if $vmStk> $summary[$pid]->{vmStkMax};
$summary[$pid]->{vmLibMin}=$vmLib if $vmLib< $summary[$pid]->{vmLibMin};
$summary[$pid]->{vmLibMax}=$vmLib if $vmLib> $summary[$pid]->{vmLibMax};
$summary[$pid]->{sysT}+=$sysT;
$summary[$pid]->{usrT}+=$usrT;
$summary[$pid]->{accumT}=$accum;
if ($processIOFlag)
{
$summary[$pid]->{RKB}+=$rkb;
$summary[$pid]->{WKB}+=$wkb;
$summary[$pid]->{RKBC}+=$rkbc;
$summary[$pid]->{WKBC}+=$wkbc;
$summary[$pid]->{RSYS}+=$rsys;
$summary[$pid]->{WSYS}+=$wsys;
$summary[$pid]->{CNCL}+=$cncl;
$summary[$pid]->{majF}+=$majF;
$summary[$pid]->{minF}+=$minF;
}
}
}
# This gets called twice! Once when we're ready to process a NEW file and
# again to write out the process summary data for the LAST log we processed
sub printProcAnalyze
{
print "Write process summary data to: $lastLogPrefix\n" if $debug & 8192;
# Note that since this is the only place we write to these files, lets open
# them here instead of trying to do it in newlog especially since newlog has
# no way of knowing if there will even be any data to write to them!
open PRCS, ">$lastLogPrefix.prcs" or
logmsg("F", "Couldn't create '$lastLogPrefix.prcs'") if !$zFlag;
$ZPRCS=Compress::Zlib::gzopen("$lastLogPrefix.prcs.gz", 'wb') or
logmsg("F", "Couldn't create '$lastLogPrefix.prcs.gz'") if $zFlag;
# NOTE - we're not printing the CPU since processes can migrate and it's not really
# meaningful yet. Perhaps someday I'll do more with it.
my $header;
$header= "Date${SEP}From${SEP}Thru${SEP}Pid${SEP}User${SEP}PPid${SEP}ExeSize${SEP}SizeMin${SEP}";
$header.="SizeMax${SEP}LckMin${SEP}LckMax${SEP}RSSMin${SEP}RSSMax${SEP}DataMin${SEP}DataMax${SEP}";
$header.="StkMin${SEP}StkMax${SEP}LibMin${SEP}LibMax${SEP}sysT${SEP}usrT${SEP}PCT${SEP}accumT${SEP}";
$header.="RKB${SEP}WKB${SEP}RKBC${SEP}WKBC${SEP}RSYS${SEP}WSYS${SEP}CNCL${SEP}" if $processIOFlag;
$header.="threadsMin${SEP}threadsMax${SEP}";
$header.="majF${SEP}minF${SEP}Command\n";
print PRCS $header if !$zFlag;
$ZPRCS->gzwrite($header) or logmsg("E", "Error writing PCRS header") if $zFlag;
my $line;
my ($date, $timefrom, $timethru);
foreach my $pid (keys %analyzed)
{
# Date always come from 'from' field
($date,$timefrom)=cvtT4($summary[$pid]->{timefrom});
# if process only ran for one interval the duration would be 0 and we can't allow that to be a divisor below.
my $pidDuration=$summary[$pid]->{timethru}-$summary[$pid]->{timefrom};
$pidDuration=1 if $pidDuration==0;
# NOTE - since sys/usr times in jiffies DON'T multiply by 100
$line=sprintf("%s$SEP%s$SEP%s$SEP%d$SEP%s$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%s$SEP%s$SEP%6.2f$SEP%s$SEP",
$date, $timefrom, (cvtT4($summary[$pid]->{timethru}))[1],
$summary[$pid]->{pid},
$summary[$pid]->{user},
$summary[$pid]->{ppid},
$summary[$pid]->{vmExe},
$summary[$pid]->{vmSizeMin},
$summary[$pid]->{vmSizeMax},
$summary[$pid]->{vmLckMin},
$summary[$pid]->{vmLckMax},
$summary[$pid]->{vmRSSMin},
$summary[$pid]->{vmRSSMax},
$summary[$pid]->{vmDataMin},
$summary[$pid]->{vmDataMax},
$summary[$pid]->{vmStkMin},
$summary[$pid]->{vmStkMax},
$summary[$pid]->{vmLibMin},
$summary[$pid]->{vmLibMax},
cvtT3($summary[$pid]->{sysT}),
cvtT3($summary[$pid]->{usrT}),
($summary[$pid]->{sysT}+$summary[$pid]->{usrT})/$pidDuration/$procAnalCounter,
$summary[$pid]->{accumT});
$line.=sprintf("%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP",
$summary[$pid]->{RKB},
$summary[$pid]->{WKB},
$summary[$pid]->{RKBC},
$summary[$pid]->{WKBC},
$summary[$pid]->{RSYS},
$summary[$pid]->{WSYS},
$summary[$pid]->{CNCL})
if $processIOFlag;
$line.=sprintf("%d$SEP%d$SEP", $summary[$pid]->{threadsMin}, $summary[$pid]->{threadsMax});
$line.=sprintf("%d$SEP%d$SEP%s\n", $summary[$pid]->{majF}, $summary[$pid]->{minF}, $summary[$pid]->{command});
print PRCS $line if !$zFlag;
$ZPRCS->gzwrite($line) or logmsg('E', "Error writing to prcs") if $zFlag;
}
undef %summary;
$procAnalyzed=0;
close PRCS;
$ZPRCS->gzclose() if $zFlag;
$procAnalCounter=0;
}
sub slabAnalyze
{
$slabAnalCounter++;
if ($slabinfoFlag)
{
for (my $i=0; $i<$slabIndexNext; $i++)
{
slabAnalyze2($slabName[$i], $slabSlabAllTotB[$i]);
}
}
else
{
foreach my $first (sort keys %slabfirst)
{
slabAnalyze2($slabfirst{$first}, $slabdata{$slab}->{total});
}
}
}
sub slabAnalyze2
{
my $name=shift;
my $size=shift;
if (!defined($slabMemTotMin{$name}))
{
$slabMemTotMin{$name}=1024*1024*1024*1024; # 1TB
$slabMemTotMax{$name}=0;
$slabMemTotFirst{$name}=$size;
}
$slabMemTotMin{$name}=$size if $size<$slabMemTotMin{$name};
$slabMemTotMax{$name}=$size if $size>$slabMemTotMax{$name};
$slabMemTotLast{$name}=$size;
}
sub printSlabAnalyze
{
print "Write slab summary data to: $lastLogPrefix\n" if $debug & 8192;
open SLBS, ">$lastLogPrefix.slbs" or
logmsg("F", "Couldn't create '$lastLogPrefix.slbs'") if !$zFlag;
$ZSLBS=Compress::Zlib::gzopen("$lastLogPrefix.slbs.gz", 'wb') or
logmsg("F", "Couldn't create '$lastLogPrefix.slbs.gz'") if $zFlag;
my $header=sprintf("%-20s %10s %10s %10s %10s %8s %8s\n",
'Slab Name', 'Start', 'End', 'Minimum', 'Maximum', 'Change', 'Pct');
print SLBS $header if !$zFlag;
$ZSLBS->gzwrite($header) or logmsg("E", "Error writing SLBS header") if $zFlag;
foreach my $name (sort keys %slabMemTotMin)
{
next if $slabMemTotMax{$name}==0;
my $diff=$slabMemTotMax{$name}-$slabMemTotMin{$name};
my $line=sprintf("%-20s %10d %10d %10d %10d %8d %8.2f\n",
$name, $slabMemTotFirst{$name}, $slabMemTotLast{$name},
$slabMemTotMin{$name}, $slabMemTotMax{$name}, $diff,
$slabMemTotMin{$name} ? 100*$diff/$slabMemTotMin{$name} : 0);
print SLBS $line if !$zFlag;
$ZSLBS->gzwrite($line) or logmsg('E', "Error writing to slbs") if $zFlag;
}
close SLBS;
$ZSLBS->gzclose() if $zFlag;
# Reset for next time
$slabAnalCounter=0;
undef %slabTotalMemLast;
undef %slabMemTotMin;
undef %slabMemTotMax;
undef %slabMemTotLast;
}
# like printPlotProc(), this only goes to .slb and we don't care about --logtoo
sub printPlotSlab
{
my $date=shift;
my $time=shift;
my ($slabHeaders, $slabPlot);
$slabHeaders='';
if (!$headersPrintedSlab)
{
$slabHeaders=$commonHeader if $logToFileFlag;
$slabHeaders.=$slubHeader if $logToFileFlag && $slubinfoFlag;
$slabHeaders.=(!$utcFlag) ? "#Date${SEP}Time" : '#UTC';
if ($slabinfoFlag)
{
$slabHeaders.="${SEP}SlabName${SEP}ObjInUse${SEP}ObjInUseB${SEP}ObjAll${SEP}ObjAllB${SEP}";
$slabHeaders.="SlabInUse${SEP}SlabInUseB${SEP}SlabAll${SEP}SlabAllB${SEP}SlabChg${SEP}SlabPct\n";
}
else
{
$slabHeaders.="${SEP}SlabName${SEP}ObjSize${SEP}ObjPerSlab${SEP}ObjInUse${SEP}ObjAvail${SEP}";
$slabHeaders.="SlabSize${SEP}SlabNumber${SEP}MemUsed${SEP}MemTotal${SEP}SlabChg${SEP}SlabPct\n";
}
$headersPrintedSlab=1;
}
my $datetime=(!$utcFlag) ? "$date$SEP$time": time;
$datetime.=".$usecs" if $options=~/m/;
$slabPlot=$slabHeaders;
# O l d S l a b F o r m a t
if ($slabinfoFlag)
{
for (my $i=0; $i{objects};
my $numSlabs= $slabdata{$slab}->{slabs};
next if ($slabOpts=~/s/ && $slabdata{$slab}->{objects}==0) ||
($slabOpts=~/S/ && $slabdata{$slab}->{lastobj}==$numObjects &&
$slabdata{$slab}->{lastslabs}==$numSlabs);
$slabPlot.=sprintf("$datetime$SEP%s$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d$SEP%d\n",
$first, $slabdata{$slab}->{slabsize}, $slabdata{$slab}->{objper},
$numObjects, $slabdata{$slab}->{avail}, ($PageSize<<$slabdata{$slab}->{order})/1024,
$numSlabs, $slabdata{$slab}->{used}/1024, $slabdata{$slab}->{total}/1024,
$slabdata{$slab}->{memchg}, $slabdata{$slab}->{mempct});
# So we can tell when something changes
$slabdata{$slab}->{lastobj}= $numObjects;
$slabdata{$slab}->{lastslabs}=$numSlabs;
}
}
# See printPlotProc() for details on this...
# Also note we're printing the whole thing in one call vs 1 call/line and we
# only want to print when there's data since filtering can result in blank
# lines. Finally, since writeData() appends a find \n, we need to strip it.
if ($slabPlot ne '')
{
$oneline='';
$slabPlot=~s/\n$//;
writeData(0, '', \$slabPlot, SLB, $ZSLB, 'slb', \$oneline);
writeData(1, '', undef, $LOG, undef, undef, \$oneline)
if !$logToFileFlag || ($sockFlag && $export eq '');
}
}
sub elanCheck
{
my $saveRails=$NumXRails;
$NumXRails=0;
if (!-e "/proc/qsnet")
{
logmsg('W', "no interconnect data found (/proc/qsnet missing)")
if $inactiveElanFlag==0;
$inactiveElanFlag=1;
}
else
{
$NumXRails++ if -e "/proc/qsnet/ep/rail0";
$NumXRails++ if -e "/proc/qsnet/ep/rail1";
# Now that I changed from using `cat` to cat(), let's just do
# this each time.
if ($NumXRails)
{
$XType='Elan';
$XVersion=cat('/proc/qsnet/ep/version');
chomp $XVersion;
}
else
{
logmsg('W', "/proc/qsnet exists but no rail stats found. is the driver loaded?")
if $inactiveElanFlag==0;
$inactiveElanFlag=1;
}
}
print "ELAN Change -- OldRails: $saveRails NewRails: $NumXRails\n"
if $debug & 2 && $NumXRails ne $saveRails;
return ($NumXRails ne $saveRails) ? 1 : 0;
}
sub ibCheck
{
my $saveHCANames=$HCANames;
my $activePorts=0;
my ($line, @lines, $port);
# Just because we have hardware doesn't mean any drivers installed and
# the assumption for now is that's the case if you can't find vstat.
# Since VStat can be a list, reset to the first that is found (if any)
$NumHCAs=0;
my $found=0;
foreach my $temp (split(/:/, $VStat))
{
if (-e $temp)
{
$found=1;
$VStat=$temp;
last;
}
}
# This error can only happen when NOT open fabric
if (!-e $SysIB && !$found)
{
logmsg('E', "Found HCA(s) but no software OR monitoring disabled in $configFile")
if $inactiveIBFlag==0;
$mellanoxFlag=0;
$inactiveIBFlag=1;
return(0);
}
# We need the names of the interfaces and port info, but it depends on the
# type of IB we're dealing with. In the case of 'vib' we get them via 'vtstat'
# and in the case of ofed via '/sys'. However, in very rare cases someone might
# have both stacks installed so just because we find 'vstat' doesn't mean vib is
# loaded.
my ($maxPorts, $numPorts)=(0,0);
$HCANames='';
if (-e $VStat)
{
@lines=`$VStat`;
foreach $line (@lines)
{
if ($line=~/hca_id=(.+)/)
{
# We need to track max ports across all HCAs. Most likely this
# is a contant.
$maxPorts=$numPorts if $numPorts>$maxPorts;
$numPorts=0;
$NumHCAs++;
$HCAName[$NumHCAs-1]=$1;
$HCAPorts[$NumHCAs-1]=0; # none active yet
$HCANames.=" $1";
}
elsif ($line=~/port=(\d+)/)
{
$port=$1;
$numPorts++;
}
elsif ($line=~/port_state=(.+)/)
{
$portState=($1 eq 'PORT_ACTIVE') ? 1 : 0;
$HCAPorts[$NumHCAs-1][$port]=$portState;
if ($portState)
{
print " VIB Port: $port\n" if $debug & 2;
$HCANames.=":$port";
$activePorts++;
}
}
$maxPorts=$numPorts if $numPorts>$maxPorts;
}
# Only if we found any HCAs (since 'vib' may not actually be loaded...)
$VoltaireStats=(-e '/proc/voltaire/adaptor-mlx/stats') ?
'/proc/voltaire/adaptor-mlx/stats' : '/proc/voltaire/ib0/stats'
if $NumHCAs;
}
# To get here, either no 'vib' OR 'vib' is there but not loaded
if ($NumHCAs==0)
{
my (@ports, $state, $file, $lid);
@lines=ls($SysIB);
foreach $line (@lines)
{
$line=~/(.*)(\d+)$/;
$devname=$1;
$devnum=$2;
# While this should work for any ofed compliant adaptor, doing it this
# way at least makes it more explicit which ones have been found to work.
if ($devname=~/mthca|mlx4_|qib/)
{
$HCAName[$NumHCAs]=$devname;
$HCAPorts[$NumHCAs]=0; # none active yet
$HCANames.=" $devname";
$file=$SysIB;
$file.="/$devname";
$file.=$devnum;
$file.="/ports";
@ports=ls($file);
$maxPorts=scalar(@ports) if scalar(@ports)>$maxPorts;
foreach $port (@ports)
{
$port=~/(\d+)/;
$port=$1;
$state=cat("$file/$1/state");
$state=~/.*: *(.+)/;
$portState=($1 eq 'ACTIVE') ? 1 : 0;
$HCAPorts[$NumHCAs][$port]=$portState;
chomp($lid=cat("$file/$port/lid"));
$HCALids[$NumHCAs][$port]=$lid;
if ($portState)
{
print " OFED Port: $port LID: $lid\n" if $debug & 2;
$HCANames.=":$port";
$activePorts++;
}
}
}
$NumHCAs++;
}
}
$HCANames=~s/^ //;
# Now we need to know port states for header.
$HCAPortStates='';
for ($i=0; $i<$NumHCAs; $i++)
{
for (my $j=1; $j<=scalar($maxPorts); $j++)
{
# The expectation is the number of ports is contant on all HCAs
# but just is case they're not, set extras to 0.
$HCAPorts[$i][$j]=0 if !defined($HCAPorts[$i][$j]);
$HCAPortStates.=$HCAPorts[$i][$j];
}
$HCAPortStates.=':';
}
$HCAPortStates=~s/:$//;
# only report inactive status once per day OR after something changed
if ($activePorts==0)
{
logmsg('E', "Found $NumHCAs HCA(s) but none had any active ports")
if $inactiveIBFlag==0;
$inactiveIBFlag=1;
}
# The names include active ports too so changes can be detected.
$changeFlag=($HCANames ne $saveHCANames) ? 1 : 0;
print "IB Change -- OldHCAs: $saveHCANames NewHCAs: $HCANames\n"
if $debug & 2 && $HCANames ne $saveHCANames;
return ($activePorts && $HCANames ne $saveHCANames) ? 1 : 0;
}
sub lustreCheckClt
{
# don't bother checking if specific services were specified and not this one
return 0 if $lustreSvcs ne '' && $lustreSvcs!~/c/i;
my ($saveFS, $saveOsts, $saveInfo, @lustreFS, @lustreDirs);
my ($dir, $dirname, $inactiveFlag);
# We're saving the info because as unlikely as it is, if the ost or fs state
# changes without their numbers changing, we need to know!
$saveFS= $NumLustreFS;
$saveOsts= $NumLustreCltOsts;
$saveInfo= $lustreCltInfo;
undef @lustreCltDirs;
undef @lustreCltFS;
undef @lustreCltFSCommon;
undef @lustreCltOsts;
undef @lustreCltOstFS;
undef @lustreCltOstDirs;
# G e t F i l e s y s t e m N a m e s
$FSWidth=0;
@lustreFS=glob("/proc/fs/lustre/llite/*");
$lustreCltInfo='';
foreach my $dir (@lustreFS)
{
# in newer versions of lustre, the fs name was dropped from uuid, so look here instead
# which does exist in earlier versions too, but we didn't look there sooner because
# uuid is still used in other cases and I wanted to be consistent.
my $commonName=cat("$dir/lov/common_name");
chomp $commonName;
my $fsName=(split(/-/, $commonName))[0];
# we use the dirname for finding 'stats' and fsname for printing.
# we may need the common name to make osts back to filesystems
my $dirname=basename($dir);
push @lustreCltDirs, $dirname;
push @lustreCltFS, $fsName;
push @lustreCltFSCommon, $commonName;
$lustreCltInfo.="$fsName: ";
$FSWidth=length($fsName) if $FSWidth$saveFS;
# O n l y F o r ' - - l u s t o p t s B / O ' G e t O S T N a m e s
undef %lustreCltOstMappings;
$inactiveFlag=0;
$NumLustreCltOsts='-'; # only meaningful for --lustopts O
if ($CltFlag && $lustOpts=~/[BO]/)
{
# we first need to get a list of all the OST uuids for all the filesystems, noting
# the 1 passed to cat() tells it to read until EOF
foreach my $commonName (@lustreCltFSCommon)
{
my $fsName=(split(/-/, $commonName))[0];
my $obds=cat("/proc/fs/lustre/lov/$commonName/target_obd", 1);
foreach my $obd (split(/\n/, $obds))
{
my ($uuid, $state)=(split(/\s+/, $obd))[1,2];
next if $state ne 'ACTIVE';
$lustreCltOstMappings{$uuid}=$fsName;
}
}
$lustreCltInfo=''; # reset by adding in OSTs
$NumLustreCltOsts=0;
@lustreDirs=glob("/proc/fs/lustre/osc/*");
foreach $dir (@lustreDirs)
{
# Since we're looking for OST subdirectories, ignore anything not a directory
# which for now is limted to 'num_refs', but who knows what the future will
# hold. As for the 'MNT' test, I think that only applied to older versions
# of lustre, certainlu tp HP-SFS.
next if !-d $dir; # currently only the 'num_refs' file
next if $cfsVersion lt '1.6.0' && $dir!~/\d+_MNT/;
# Looks like if you're on a 1.6.4.3 system (and perhaps earlier) that is both
# a client as well as an MDS, you'll see MDS specific directories with names
# like - lustre-OST0000-osc, whereas lustre-OST0000-osc-000001012e950400 is the
# client directory we want, so...
next if $dir=~/\-osc$/;
# if ost closed (this happens when new filesystems get created), ignore it.
# note that newer versions of lustre added a sstate and sets it to DEACTIVATED
my ($uuid, $state,$sstate)=split(/\s+/, cat("$dir/ost_server_uuid"));
next if $state=~/CLOSED|DISCONN/ || $sstate=~/DEACT/;
# uuids look something like 'xxx-ost_UUID' and you can actully have a - or _
# following the xxx so drop the beginning/end this way in case an embedded _
# in ost name itself.
$ostName=$uuid;
$ostName=~s/.*?[-_](.*)_UUID/$1/;
$fsName=$lustreCltOstMappings{$uuid};
$OstWidth=length($ostName) if $OstWidth$saveOsts;
}
$lustreCltInfo=~s/ $//;
# Change info is important even when not logging except during initialization
if ($lustreCltInfo ne $saveInfo)
{
my $comment=($filename eq '') ? '#' : '';
my $text="Lustre CLT OSTs Changed -- Old: $saveInfo New: $lustreCltInfo";
logmsg('W', "${comment}$text") if !$firstPass;
print "$text\n" if $firstPass && $debug & 8;
}
return ($lustreCltInfo ne $saveInfo) ? 1 : 0;
}
sub lustreCheckMds
{
# don't bother checking if specific services were specified and not this one
return 0 if $lustreSvcs ne '' && $lustreSvcs!~/m/i;
# if this wasn't an MDS and still isn't, nothing has changed
my $type=($cfsVersion lt '1.6.0') ? 'MDT' : 'MDS';
return 0 if !$NumMds && !-e "/proc/fs/lustre/mdt/$type/mds/stats";
my ($saveMdsNames, @mdsDirs, $mdsName);
$saveMdsNames=$MdsNames;
$MdsNames='';
$NumMds=$MdsFlag=0;
@mdsDirs=glob("/proc/fs/lustre/mds/*");
foreach $mdsName (@mdsDirs)
{
next if $mdsName=~/num_refs/;
$mdsName=basename($mdsName);
$MdsNames.="$mdsName ";
$NumMds++;
$MdsFlag=1; # for consistency with CltFlag and OstFlag
}
$MdsNames=~s/ $//;
# Change info is important even when not logging except during initialization
if ($MdsNames ne $saveMdsNames)
{
my $comment=($filename eq '') ? '#' : '';
my $text="Lustre MDS FS Changed -- Old: $saveMdsNames New: $MdsNames";
logmsg('W', "${comment}$text") if !$firstPass;
print "$text\n" if $firstPass && $debug & 8;
}
return ($MdsNames ne $saveMdsNames) ? 1 : 0;
}
sub lustreCheckOst
{
# don't bother checking if specific services were specified and not this one
return 0 if $lustreSvcs ne '' && $lustreSvcs!~/o/i;
# if this wasn't an OST and still isn't, nothing has changed.
return 0 if !$NumOst && !-e "/proc/fs/lustre/obdfilter";
my ($saveOst, $saveOstNames, @ostFiles, $file, $ostName, $subdir);
$saveOst=$NumOst;
$saveOstNames=$OstNames;
undef @lustreOstSubdirs;
# check for OST files
$OstNames='';
$NumOst=$OstFlag=0;
@ostFiles=glob("/proc/fs/lustre/obdfilter/*/stats");
foreach $file (@ostFiles)
{
$file=~m[/proc/fs/lustre/obdfilter/(.*)/stats];
$subdir=$1;
push @lustreOstSubdirs, $subdir;
$temp=cat("/proc/fs/lustre/obdfilter/$subdir/uuid");
$ostName=transLustreUUID($temp);
$OstWidth=length($ostName) if $OstWidth$saveOst;
# Change info is important even when not logging except during initialization
if ($OstNames ne $saveOstNames)
{
my $comment=($filename eq '') ? '#' : '';
my $text="Lustre OSS OSTs Changed -- Old: $saveOstNames New: $OstNames";
logmsg('W', "${comment}$text") if !$firstPass;
print "$text\n" if $firstPass && $debug & 8;
}
return ($OstNames ne $saveOstNames) ? 1 : 0;
}
sub transLustreUUID
{
my $name=shift;
my $hostRoot;
# This handles names like OST_Lustre9_2_UUID or OST_Lustre9_UUID or in
# the case of SFS something like ost123_UUID, changing them to just 0,9
# or ost123.
chomp $name;
$hostRoot=$Host;
$hostRoot=~s/\d+$//;
$name=~s/OST_$hostRoot\d+//;
$name=~s/_UUID//;
$name=~s/_//;
$name=0 if $name eq '';
return($name);
}
# since it seems OFED changes the locations of perfquery and ofed_info
# with each release, we're gonna check for them here and if we can't find
# them, do an 'rpm -qal' and look for them there and on finding them,
# update /etc/collectl.conf (if we can)
sub getOfedPath
{
my $list= shift;
my $name= shift;
my $label=shift;
my $found='';
foreach my $path (split(/:/, $list))
{
if (-e $path)
{
$found=$path;
last;
}
}
# RHEL54 stopped shipping it so we need to know RH version first
my $RHVersion=($Distro=~/Red Hat.*(\d+\.\d+)/) ? $1 : '';
# Can't find in standard places so ask rpm, but only if it's there
if ($found eq '' && -e $Rpm && $RHVersion ne '' && $RHVersion<5.4)
{
# This is something we really don't want to have to be doing
logmsg('W', "Cannot find '$name' in ${configFile}'s OFED search list, checking with rpm");
$command="$Rpm -qal | $Grep $name | $Grep -v man";
print "Command: $command\n" if $debug & 2;
$found=`$command`;
if ($found ne '')
{
if (-w $configFile)
{
chomp($found);
logmsg('I', "Adding '$found' to '$label' in $configFile");
my $conf=`$Cat $configFile`;
$conf=~s/($label\s+=\s+)(.*)$/$1$found:$2/m;
open CONF, ">$configFile" or logmsg("F", "Couldn't write to $configFile so do it manually!");
print CONF $conf;
close CONF;
}
else
{
logmsg('W', "found '$name' in rpm but $configFile not writeable so not updated");
}
}
}
return($found);
}
# While tempted to put this in collectl main line, this is really only used during formatting
sub loadEnvRules
{
my $envStdFlag=($envRules eq '') ? 1 : 0;
my $ruleFile=($envStdFlag) ? "$ReqDir${Sep}envrules.std" : $envRules;
open TMP, "<$ruleFile" or logmsg('F', "Cannot open '$ruleFile'");
my $skipFlag=1 if $envStdFlag; # if 'std', need to find right stanza
my ($index, $type);
while (my $line=)
{
next if $line=~/^#|^\s*$/;
chomp $line;
if ($line=~/>(.*){$type}->[$index]->{f1}=$f1;
$ipmiFile->{$type}->[$index]->{f2}=$f2;
$index++;
}
close TMP;
}
##################################################
# These are MUCH faster than the linux commands
# since we don't have to start a new process!
##################################################
sub cat
{
my $file=shift;
my $eof= shift;
my $temp;
if (!open CAT, "<$file")
{
logmsg("W", "Can't open '$file'");
$temp='';
}
else
{
# if 'eof' set, return entire file, otherwise just 1st line.
while (my $line=)
{
$temp.=$line;
last if !defined($eof);
}
close CAT;
}
return($temp);
}
sub ls
{
my @dirs;
opendir DIR, $_[0];
while (my $line=readdir(DIR))
{
next if $line=~/^\./;
push @dirs, $line;
}
close DIR;
return(@dirs);
}
1;
collectl-3.6.9/lexpr.ph 0000644 0001750 0001750 00000060431 12230241726 013164 0 ustar mjs mjs # copyright, 2003-2009 Hewlett-Packard Development Company, LP
# Call with --custom "lexpr[,switches]
# Debug
# 1 - show all names/values, noting the timestamps are now!
# 2 - just show names/values 'sent'
# 4 - include real times in timestamps (useful for testing) along with skipped intervals
# 8 - do not send anything
# (useful when displaying normal output on terminal)
# 16 - show 'x' processing
our $lexInterval;
my ($lexSubsys, $lexDebug, $lexCOFlag, $lexTTL, $lexFilename, $lexSumFlag);
my (%lexDataLast, %lexDataMin, %lexDataMax, %lexDataTot, %lexTTL);
my ($lexColInt, $lexSendCount, $lexFlags);
my ($lexMinFlag, $lexMaxFlag, $lexAvgFlag, $lexTotFlag)=(0,0,0,0);
my $lexOneTB=1024*1024*1024*1024;
my $lexSamples=0;
my $lexOutputFlag=1;
my $lexFirstInt=1;
my $lexAlignFlag=0;
my $lexCounter;
my $lexExtName='';
sub lexprInit
{
error('--showcolheader not supported by lexpr') if $showColFlag;
# on the odd chance someone did -s-all and have other ways to generate data, collectl
# hasn't yet hit the code that resets $subsys so we have to do it here.
$subsys='' if $userSubsys eq '-all';
# Defaults for options
$lexDebug=$lexCOFlag=0;
$lexFilename='';
$lexInterval='';
$lexSubsys=$subsys;
$lexTTL=5;
foreach my $option (@_)
{
my ($name, $value)=split(/=/, $option, 2); # in case more than 1 = in single option string
error("invalid lexpr option '$name'") if $name!~/^[dfhisx]?$|^align$|^co$|^ttl$|^min$|^max$|^avg$|^tot$/;
$lexAlignFlag=1 if $name eq 'align';
$lexCOFlag=1 if $name eq 'co';
$lexDebug=$value if $name eq 'd';
$lexFilename=$value if $name eq 'f';
$lexInterval=$value if $name eq 'i';
$lexSubsys=$value if $name eq 's';
$lexExtName=$value if $name eq 'x';
$lexTTL=$value if $name eq 'ttl';
$lexMinFlag=1 if $name eq 'min';
$lexMaxFlag=1 if $name eq 'max';
$lexAvgFlag=1 if $name eq 'avg';
$lexTotFlag=1 if $name eq 'tot';
help() if $name eq 'h';
last if $lexExtName ne '';
}
# If importing data, and if not reporting anything else, $subsys will be ''
$lexSumFlag=$lexSubsys=~/[cdfilmnstxE]/ ? 1 : 0;
# s= disables ALL subsys, only makes sense with imports
error("lexpr subsys options '$lexSubsys' not a proper subset of '$subsys'")
if $subsys ne '' && $lexSubsys ne '' && $lexSubsys!~/^[$subsys]+$/;
error("lexpr cannot write a snapshot file and use a socket at the same time")
if $sockFlag && $lexFilename ne '';
# Using -f and f= will not result in raw or plot file so need this message.
error ("using lexpr option 'f=' AND -f requires -P and/or --rawtoo")
if $lexFilename ne '' && $filename ne '' && !$plotFlag && !$rawtooFlag;
# if -f, use that dirname/L for snampshot file; otherwise use f= for it.
$lexFilename=(-d $filename) ? "$filename/L" : dirname($filename)."/L"
if $lexFilename eq '' && $filename ne '';
$lexFlags=$lexMinFlag+$lexMaxFlag+$lexAvgFlag|$lexTotFlag;
error("only 1 of 'min', 'max', 'avg' or 'tot' with 'lexpr'") if $lexFlags>1;
# check for consistent intervals in interactive mode
if ($playback eq '')
{
$lexColInt=(split(/:/, $interval))[0];
$lexInterval=$lexColInt if $lexInterval eq '';
$lexSendCount=int($lexInterval/$lexColInt);
error("lexpr interval of '$lexInterval' is not a multiple of collectl interval of '$lexColInt' seconds")
if $lexColInt*$lexSendCount != $lexInterval;
error("'min', 'max', 'avg' & 'tot' require lexpr 'i' that is > collectl's -i")
if $lexFlags && $lexSendCount==1;
if ($lexAlignFlag)
{
my $div1=int(60/$lexColInt);
my $div2=int($lexColInt/60);
error("'align' requires collectl interval be a factor or multiple of 60 seconds")
if ($lexColInt<=60 && $div1*$lexColInt!=60) || ($lexColInt>60 && $div2*60!=$lexColInt);
error("'align' only makes sense when multiple samples/interval") if $lexInterval<=$lexColInt;
error("'lexpr,align' requires -D or --align") if !$alignFlag && !$daemonFlag;
}
}
if ($lexExtName ne '')
{
# build up swiches from EVERYTHING seen after x=
my $xSeen=0;
my $switches='';
foreach my $option (@_)
{
$xSeen=1 if $option=~/^x/;
$switches.="$option," if $xSeen && $option!~/^x/;
}
$switches=~s/,$//;
($lexExtName, $switches)=(split(/:/, $lexExtName, 2))[0,1] if $lexExtName=~/:/; # backwards compatibility with : for switches
$lexExtBase=$lexExtName;
$lexExtBase=~s/\..*//; # in case extension
$lexExtName.='.ph' if $lexExtName!~/\./;
#print "NAME: $lexExtName Switches: $switches\n";
$tempName=$lexExtName; # name for error message before prepending with directory
$lexExtName="$ReqDir/$lexExtName" if !-e $lexExtName;
if (!-e "$lexExtName")
{
my $temp="can't find lexpr extension file '$tempName' in ./";
$temp.=" OR $ReqDir/" if $ReqDir ne '.';
error($temp);
}
require $lexExtName;
print "$lexExtName loaded\n" if $lexDebug & 16;
# rather than pass an undefined switch, if not there don't pass anything
my $initName="${lexExtBase}Init";
if (defined($switches))
{
print "$initName($switches)\n" if $lexDebug & 16;
&$initName($switches);
}
else
{ &$initName(); }
}
# need to reset here in case processing multiple files
$lexCounter=0;
}
sub lexpr
{
# since our init routine gets call BEFORE playback processing we have to wait until first interval to do this
if ($lexFirstInt && $playback ne '')
{
# you might be able to align with data collected with --align or -D, but I'd rather discourage this
$lexColInt=(split(/:/, $recInterval))[0];
$lexInterval=$lexColInt if $lexInterval eq '';
$lexSendCount=int($lexInterval/$lexColInt);
error("lexpr interval of '$lexInterval' is not a multiple of recorded interval of '$lexColInt' seconds")
if $lexColInt*$lexSendCount != $lexInterval;
error("'align' not supported with -p") if $lexAlignFlag;
error("'min', 'max', 'avg' & 'tot' require lexpr 'i' that is > collectl's -i")
if $lexFlags && $lexSendCount==1;
}
$lexFirstInt=0;
# if not time to print and we're not doing min/max/avg/tot, there's nothing to do.
# BUT if align, always make sure time aligns to top of minute based on i= and NOT sendCount
$lexCounter++;
$lexSamples++;
$lexOutputFlag=(($lexCounter % $lexSendCount) ==0) ? 1 : 0 if !$lexAlignFlag;
$lexOutputFlag=(!(int($lastSecs[$rawPFlag]) % $lexInterval)) ? 1 : 0 if $lexAlignFlag;
#print "Align: $lexAlignFlag Counter: $lexCounter LexSend: $lexSendCount Last: $lastSecs[$rawPFlag] Output: $lexOutputFlag\n";
return if (!$lexOutputFlag && $lexFlags==0);
my ($cpuSumString,$cpuDetString)=('','');
if ($lexSubsys=~/c/i)
{
if ($lexSubsys=~/c/)
{
# CPU utilization is a % and we don't want to report fractions
my $i=$NumCpus;
$cpuSumString.=sendData("cputotals.num", $i, 1);
$cpuSumString.=sendData("cputotals.user", $userP[$i], 1);
$cpuSumString.=sendData("cputotals.nice", $niceP[$i], 1);
$cpuSumString.=sendData("cputotals.sys", $sysP[$i], 1);
$cpuSumString.=sendData("cputotals.wait", $waitP[$i], 1);
$cpuSumString.=sendData("cputotals.irq", $irqP[$i], 1);
$cpuSumString.=sendData("cputotals.soft", $softP[$i], 1);
$cpuSumString.=sendData("cputotals.steal", $stealP[$i], 1);
$cpuSumString.=sendData("cputotals.idle", $idleP[$i], 1);
# These 2 are redundant, but also handy
$cpuSumString.=sendData("cputotals.systot", $sysP[$i]+$irqP[$i]+$softP[$i]+$stealP[$i], 1);
$cpuSumString.=sendData("cputotals.usertot", $userP[$i]+$niceP[$i], 1);
$cpuSumString.=sendData("cputotals.total", $sysP[$i]+$irqP[$i]+$softP[$i]+$stealP[$i]+$userP[$i]+$niceP[$i], 1);
$cpuSumString.=sendData("ctxint.ctx", $ctxt/$intSecs);
$cpuSumString.=sendData("ctxint.int", $intrpt/$intSecs);
$cpuSumString.=sendData("proc.creates", $proc/$intSecs);
$cpuSumString.=sendData("proc.runq", $loadQue, 1);
$cpuSumString.=sendData("proc.run", $loadRun, 1);
$cpuSumString.=sendData("cpuload.avg1", $loadAvg1, 1, '%4.2f');
$cpuSumString.=sendData("cpuload.avg5", $loadAvg5, 1, '%4.2f');
$cpuSumString.=sendData("cpuload.avg15", $loadAvg15, 1,'%4.2f');
}
if ($lexSubsys=~/C/)
{
for (my $i=0; $i<$NumCpus; $i++)
{
$cpuDetString.=sendData("cpuinfo.user.cpu$i", $userP[$i], 1);
$cpuDetString.=sendData("cpuinfo.nice.cpu$i", $niceP[$i], 1);
$cpuDetString.=sendData("cpuinfo.sys.cpu$i", $sysP[$i], 1);
$cpuDetString.=sendData("cpuinfo.wait.cpu$i", $waitP[$i], 1);
$cpuDetString.=sendData("cpuinfo.irq.cpu$i", $irqP[$i], 1);
$cpuDetString.=sendData("cpuinfo.soft.cpu$i", $softP[$i], 1);
$cpuDetString.=sendData("cpuinfo.steal.cpu$i", $stealP[$i], 1);
$cpuDetString.=sendData("cpuinfo.idle.cpu$i", $idleP[$i], 1);
$cpuDetString.=sendData("cpuinfo.intrpt.cpu$i", $intrptTot[$i], 1);
$cpuSumString.=sendData("cputotals.systot.cpu$i", $sysP[$i]+$irqP[$i]+$softP[$i]+$stealP[$i], 1);
$cpuSumString.=sendData("cputotals.usertot.cpu$i", $userP[$i]+$niceP[$i], 1);
}
}
}
my ($diskSumString,$diskDetString)=('','');
if ($lexSubsys=~/d/i)
{
if ($lexSubsys=~/d/)
{
$diskSumString.=sendData("disktotals.reads", $dskReadTot/$intSecs);
$diskSumString.=sendData("disktotals.readkbs", $dskReadKBTot/$intSecs);
$diskSumString.=sendData("disktotals.writes", $dskWriteTot/$intSecs);
$diskSumString.=sendData("disktotals.writekbs", $dskWriteKBTot/$intSecs);
}
if ($lexSubsys=~/D/)
{
for (my $i=0; $i<@dskOrder; $i++)
{
# preserve display order but skip any disks not seen this interval
$dskName=$dskOrder[$i];
next if !defined($dskSeen[$i]);
next if ($dskFiltKeep eq '' && $dskName=~/$dskFiltIgnore/) || ($dskFiltKeep ne '' && $dskName!~/$dskFiltKeep/);
$diskDetString.=sendData("diskinfo.reads.$dskName", $dskRead[$i]/$intSecs);
$diskDetString.=sendData("diskinfo.readkbs.$dskName", $dskReadKB[$i]/$intSecs);
$diskDetString.=sendData("diskinfo.writes.$dskName", $dskWrite[$i]/$intSecs);
$diskDetString.=sendData("diskinfo.writekbs.$dskName", $dskWriteKB[$i]/$intSecs);
$diskDetString.=sendData("diskinfo.quelen.$dskName", $dskQueLen[$i]/$intSecs);
$diskDetString.=sendData("diskinfo.wait.$dskName", $dskWait[$i]/$intSecs);
$diskDetString.=sendData("diskinfo.svctime.$dskName", $dskSvcTime[$i]/$intSecs);
$diskDetString.=sendData("diskinfo.util.$dskName", $dskUtil[$i]/$intSecs);
}
}
}
my $nfsString='';
if ($lexSubsys=~/f/)
{
if ($nfsSFlag)
{
$nfsString.=sendData("nfsinfo.Sread", $nfsSReadsTot/$intSecs);
$nfsString.=sendData("nfsinfo.Swrite", $nfsSWritesTot/$intSecs);
$nfsString.=sendData("nfsinfo.Smeta", $nfsSMetaTot/$intSecs);
$nfsString.=sendData("nfsinfo.Scommit",$nfsSCommitTot/$intSecs);
}
if ($nfsCFlag)
{
$nfsString.=sendData("nfsinfo.Cread", $nfsCReadsTot/$intSecs);
$nfsString.=sendData("nfsinfo.Cwrite", $nfsCWritesTot/$intSecs);
$nfsString.=sendData("nfsinfo.Cmeta", $nfsCMetaTot/$intSecs);
$nfsString.=sendData("nfsinfo.Ccommit",$nfsCCommitTot/$intSecs);
}
}
my $inodeString='';
if ($lexSubsys=~/i/)
{
$inodeString.=sendData("inodeinfo.dentrynum", $dentryNum, 1);
$inodeString.=sendData("inodeinfo.dentryunused", $dentryUnused, 1);
$inodeString.=sendData("inodeinfo.filesalloc", $filesAlloc, 1);
$inodeString.=sendData("inodeinfo.filesmax", $filesMax, 1);
$inodeString.=sendData("inodeinfo.inodeused", $inodeUsed, 1);
}
# No lustre details, at least not for now...
my $lusSumString='';
if ($lexSubsys=~/l/)
{
if ($CltFlag)
{
$lusSumString.=sendData("lusclt.reads", $lustreCltReadTot/$intSecs);
$lusSumString.=sendData("lusclt.readkbs", $lustreCltReadKBTot/$intSecs);
$lusSumString.=sendData("lusclt.writes", $lustreCltWriteTot/$intSecs);
$lusSumString.=sendData("lusclt.writekbs", $lustreCltWriteKBTot/$intSecs);
$lusSumString.=sendData("lusclt.numfs", $NumLustreFS, 1);
}
if ($MdsFlag)
{
my $getattrPlus=$lustreMdsGetattr+$lustreMdsGetattrLock+$lustreMdsGetxattr;
my $setattrPlus=$lustreMdsReintSetattr+$lustreMdsSetxattr;
my $varName=($cfsVersion lt '1.6.5') ? 'reint' : 'unlink';
my $varVal= ($cfsVersion lt '1.6.5') ? $lustreMdsReint : $lustreMdsReintUnlink;
$lusSumString.=sendData('lusmds.gattrP', $getattrPlus/$intSecs);
$lusSumString.=sendData('lusmds.sattrP', $setattrPlus/$intSecs);
$lusSumString.=sendData('lusmds.sync', $lustreMdsSync/$intSecs);
$lusSumString.=sendData("lusmds.$varName", $varVal/$intSecs);
}
if ($OstFlag)
{
$lusSumString.=sendData("lusost.reads", $lustreReadOpsTot/$intSecs);
$lusSumString.=sendData("lusost.readkbs", $lustreReadKBytesTot/$intSecs);
$lusSumString.=sendData("lusost.writes", $lustreWriteOpsTot/$intSecs);
$lusSumString.=sendData("lusost.writekbs", $lustreWriteKBytesTot/$intSecs);
}
}
my ($memString, $memDetString)=('','');
if ($lexSubsys=~/m/i)
{
if ($lexSubsys=~/m/)
{
$memString.=sendData("meminfo.tot", $memTot, 1);
$memString.=sendData("meminfo.used", $memUsed, 1);
$memString.=sendData("meminfo.free", $memFree, 1);
$memString.=sendData("meminfo.shared", $memShared, 1);
$memString.=sendData("meminfo.buf", $memBuf, 1);
$memString.=sendData("meminfo.cached", $memCached, 1);
$memString.=sendData("meminfo.slab", $memSlab, 1);
$memString.=sendData("meminfo.map", $memMap, 1);
$memString.=sendData("meminfo.anon", $memAnon, 1);
$memString.=sendData("meminfo.dirty", $memDirty, 1);
$memString.=sendData("meminfo.locked", $memLocked, 1);
$memString.=sendData("meminfo.inactive", $memInact, 1);
$memString.=sendData("meminfo.hugetot", $memHugeTot, 1);
$memString.=sendData("meminfo.hugefree", $memHugeFree, 1);
$memString.=sendData("meminfo.hugersvd", $memHugeRsvd, 1);
$memString.=sendData("meminfo.sunreclaim", $memSUnreclaim, 1);
$memString.=sendData("swapinfo.total", $swapTotal, 1);
$memString.=sendData("swapinfo.free", $swapFree, 1);
$memString.=sendData("swapinfo.used", $swapUsed, 1);
$memString.=sendData("swapinfo.in", $swapin/$intSecs);
$memString.=sendData("swapinfo.out", $swapout/$intSecs);
$memString.=sendData("pageinfo.fault", $pagefault/$intSecs);
$memString.=sendData("pageinfo.majfault", $pagemajfault/$intSecs);
$memString.=sendData("pageinfo.in", $pagein/$intSecs);
$memString.=sendData("pageinfo.out", $pageout/$intSecs);
}
if ($lexSubsys=~/M/)
{
for (my $i=0; $i<$CpuNodes; $i++)
{
foreach my $field ('used', 'free', 'slab', 'map', 'anon', 'lock', 'act', 'inact')
{
$memDetString.=sendData("numainfo.$field.$i", $numaMem[$i]->{$field}, 1);
}
}
}
}
my ($netSumString,$netDetString)=('','');
if ($lexSubsys=~/n/i)
{
if ($lexSubsys=~/n/)
{
$netSumString.=sendData("nettotals.kbin", $netRxKBTot/$intSecs);
$netSumString.=sendData("nettotals.pktin", $netRxPktTot/$intSecs);
$netSumString.=sendData("nettotals.kbout", $netTxKBTot/$intSecs);
$netSumString.=sendData("nettotals.pktout", $netTxPktTot/$intSecs);
}
if ($lexSubsys=~/N/)
{
for ($i=0; $i<@netOrder; $i++)
{
$netName=$netOrder[$i];
next if !defined($netSeen[$i]);
next if ($netFiltKeep eq '' && $netName=~/$netFiltIgnore/) || ($netFiltKeep ne '' && $netName!~/$netFiltKeep/);
next if $netName=~/lo|sit/;
$netDetString.=sendData("netinfo.kbin.$netName", $netRxKB[$i]/$intSecs);
$netDetString.=sendData("netinfo.pktin.$netName", $netRxPkt[$i]/$intSecs);
$netDetString.=sendData("netinfo.kbout.$netName", $netTxKB[$i]/$intSecs);
$netDetString.=sendData("netinfo.pktout.$netName", $netTxPkt[$i]/$intSecs);
}
}
}
my $sockString='';
if ($lexSubsys=~/s/)
{
$sockString.=sendData("sockinfo.used", $sockUsed, 1);
$sockString.=sendData("sockinfo.tcp", $sockTcp, 1);
$sockString.=sendData("sockinfo.orphan", $sockOrphan, 1);
$sockString.=sendData("sockinfo.tw", $sockTw, 1);
$sockString.=sendData("sockinfo.alloc", $sockAlloc, 1);
$sockString.=sendData("sockinfo.mem", $sockMem, 1);
$sockString.=sendData("sockinfo.udp", $sockUdp, 1);
$sockString.=sendData("sockinfo.raw", $sockRaw, 1);
$sockString.=sendData("sockinfo.frag", $sockFrag, 1);
$sockString.=sendData("sockinfo.fragm", $sockFragM, 1);
}
my $tcpString='';
if ($lexSubsys=~/t/)
{
$tcpString.=sendData("tcpinfo.iperrs", $ipErrors/$intSecs) if $tcpFilt=~/i/;
$tcpString.=sendData("tcpinfo.tcperrs", $tcpErrors/$intSecs) if $tcpFilt=~/t/;
$tcpString.=sendData("tcpinfo.udperrs", $udpErrors/$intSecs) if $tcpFilt=~/u/;
$tcpString.=sendData("tcpinfo.icmperrs", $icmpErrors/$intSecs) if $tcpFilt=~/c/;
$tcpString.=sendData("tcpinfo.tcpxerrs", $tcpExErrors/$intSecs) if $tcpFilt=~/T/;
}
my $intString='';
if ($lexSubsys=~/x/i)
{
if ($NumXRails)
{
$kbInT= $elanRxKBTot;
$pktInT= $elanRxTot;
$kbOutT= $elanTxKBTot;
$pktOutT=$elanTxTot;
}
if ($NumHCAs)
{
$kbInT= $ibRxKBTot;
$pktInT= $ibRxTot;
$kbOutT= $ibTxKBTot;
$pktOutT=$ibTxTot;
}
$intString.=sendData("iconnect.kbin", $kbInT/$intSecs);
$intString.=sendData("iconnect.pktin", $pktInT/$intSecs);
$intString.=sendData("iconnect.kbout", $kbOutT/$intSecs);
$intString.=sendData("iconnect.pktout", $pktOutT/$intSecs);
}
my $envString='';
if ($lexSubsys=~/E/i)
{
foreach $key (sort keys %$ipmiData)
{
for (my $i=0; $i{$key}}); $i++)
{
my $name=$ipmiData->{$key}->[$i]->{name};
my $inst=($key!~/power/ && $ipmiData->{$key}->[$i]->{inst} ne '-1') ? $ipmiData->{$key}->[$i]->{inst} : '';
$envString.=sendData("env.$name$inst", $ipmiData->{$key}->[$i]->{value}, 1, '%s');
}
}
}
# if any imported data, it may want to include lexpr output AND we do a little more work to
# separate the summary from the detail. also, in case any variables are gauges and we're doing
# totals we'll need to know that too.
my (@nameS, @valS, @nameD, @valD, @gaugeS, @gaugeD);
my ($impSumString, $impDetString)=('','');
for (my $i=0; $i<$impNumMods; $i++) { &{$impPrintExport[$i]}('l', \@nameS, \@valS, \@nameD, \@valD, \@gaugeS, \@gaugeD); }
foreach (my $i=0; $i$lexFilename" or logmsg("F", "Couldn't create '$lexFilename'");
print EXP $lexprRec;
close EXP;
}
$lexSamples=0;
}
# this code tightly synchronized with gexpr and graphite
sub sendData
{
my $name= shift;
my $value= shift;
my $gauge= shift;
my $format=shift;
#print "Name: $name VAL: $value\n";
# These are only undefined the very first time
if (!defined($lexTTL{$name}))
{
$lexTTL{$name}=$lexTTL;
$lexDataLast{$name}=-1;
}
# As a minor optimization, only do this when dealing with min/max/avg/tot values
if ($lexFlags)
{
# And while this should be done in init(), we really don't know how may indexes
# there are until our first pass through...
if ($lexSamples==1)
{
$lexDataMin{$name}=$lexOneTB;
$lexDataMax{$name}=0;
$lexDataTot{$name}=0;
}
$lexDataMin{$name}=$value if $lexMinFlag && $value<$lexDataMin{$name};
$lexDataMax{$name}=$value if $lexMaxFlag && $value>$lexDataMax{$name};
$lexDataTot{$name}+=$value if $lexAvgFlag;
# totals are a little different. In the case of rates, we need to multiply by the collectl
# interval to get the interval total, but for gauges we're really only doing averages
$lexDataTot{$name}+=(!$gauge) ? $value*$lexColInt : $value if $lexTotFlag;
}
return('') if !$lexOutputFlag;
# A c t u a l S e n d H a p p e n s H e r e
# If doing min/max/avg, reset $value
if ($lexFlags)
{
$value=$lexDataMin{$name} if $lexMinFlag;
$value=$lexDataMax{$name} if $lexMaxFlag;
$value=$lexDataTot{$name} if $lexTotFlag;
$value=$lexDataTot{$name}/$lexSamples if $lexAvgFlag || defined($gauge); # gauges are reported as averages
}
# Always send send data if not CO mode, but if so only send when it has
# indeed changed OR TTL about to expire
my $valSentFlag=0;
my $returnString='';
if (!$lexCOFlag || $value!=$lexDataLast{$name} || $lexTTL{$name}==1)
{
$valSentFlag=1;
$format='%d' if !defined($format);
$value+=.5 if $format=~/d/;
$returnString=sprintf("%s $format\n", $name, $value) unless $lexDebug & 8;
$lexDataLast{$name}=$value;
}
# A fair chunk of work, but worth it
if ($lexDebug & 3)
{
my ($intSeconds, $intUsecs);
if ($hiResFlag)
{
# we have to fully qualify name because or 'require' vs 'use'
($intSeconds, $intUsecs)=Time::HiRes::gettimeofday();
}
else
{
$intSeconds=time;
$intUsecs=0;
}
$intUsecs=sprintf("%06d", $intUsecs);
my ($sec, $min, $hour)=localtime($intSeconds);
my $timestamp=sprintf("%02d:%02d:%02d.%s", $hour, $min, $sec, substr($intUsecs, 0, 3));
printf "$timestamp Name: %-20s Val: %8d TTL: %d %s\n",
$name, $value, $lexTTL{$name}, ($valSentFlag) ? 'sent' : ''
if $lexDebug & 1 || $valSentFlag;
}
# TTL only applies when in 'CO' mode, noting we already made expiration
# decision above when we saw counter of 1
if ($lexCOFlag)
{
$lexTTL{$name}-- if !$valSentFlag;
$lexTTL{$name}=$lexTTL if $valSentFlag || $lexTTL{$name}==0;
}
return($returnString);
}
sub help
{
my $text=<$outfile" or error("Couldn't create '$outfile'");
print "Creating: $outfile\n";
my $state=0;
my $header='';
while (my $line=)
{
if ($line=~/^#Date/)
{
cvtHeader(\$header);
$state=1;
$line=~s/ /,/g;
$line=~s/#Date,Time(.*),?$/Sample Time$1/; # also get rid of optional trailing comma!
print OUT $line;
next;
}
if ($state==0)
{
$header.=$line;
next;
}
my ($date, $time, $therest)=split(/ /, $line, 3);
my $year=substr($date, 0, 4);
my $mon= substr($date, 4, 2);
my $day=substr($date, 6, 2);
my $month=substr('JanFebMarAprMayJunJulAugSepOctNovDec', ($mon-1)*3, 3);
my $newdate="$day-$month-$year";
$therest=~s/ /,/g;
$therest=~s/[a-z].*?,/0,/g; # for DSK and NET files, this will convert instance names to 0's
print OUT "$newdate $time,$therest";
}
}
print "skipped $skipped file(s) that did not extension(s): tab,cpu,dsk,net,nfs\n" if $skipped;
sub cvtHeader
{
my $hdrref=shift;
$$hdrref=~s/##.*//g;
$$hdrref=~s/#\s+(.*)/"$1",/g;
$$hdrref=~s/[\n\r]+//g;
$$hdrref=~s/,$//;
$$hdrref=~/Host:\s+(\S+)/;
$$hdrref="$1,$$hdrref\n";
print OUT $$hdrref;
}
sub error
{
print "$_[0]\n";
exit;
}
