1. Introduction
django-redis is a BSD Licensed, full featured Redis cache/session backend for Django.
1.1. Why use django-redis?
Because:
-
In active development.
-
Uses native redis-py url notation connection strings.
-
Modular client system (pluggable clients).
-
Master-Slave support in the default client.
-
Complete battery of tests.
-
Used in production in several projects as cache and session storage.
-
Supports infinite timeouts.
-
Facilities for raw access to Redis client/connection pool.
-
Highly configurable (can emulate memcached exception behavior, for example).
-
Unix sockets supported by default.
-
Pluggable parsers.
-
With support for python 2.7, 3.3 and 3.4
1.2. Supported django-redis versions
-
Development version: 4.0.0
-
Supported stable version: 3.8.4
1.3. How version number is handled
Versions like 3.6, 3.7, … are considered major releases and can contain some backward incompatibilities. For more information is very recommended see the changelog before update.
Versions like 3.7.0, 3.7.1, … are considered minor or bug fix releases and are should contain only bug fixes. No new features.
1.4. Requirements
1.4.1. Django version support
-
django-redis 3.8.x will maintain support for django 1.4, 1.5, 1.6, 1.7 (and maybe 1.8)
-
django-redis 4.0.0 will maintain support for django 1.6, 1.7 and 1.8
1.4.2. Redis Server Support
-
django-redis 3.8.x will maintain support for redis-server 2.6.x and upper.
-
django-redis 4.0.0 will maintain support for redis-server 2.8.x and upper.
1.4.3. Other requirements
All supported versions of django-redis depends on redis-py >= 2.10.0
.
2. User guide
2.1. Installation
The simplest way to use django-redis in your project is to install it with pip:
pip install django-redis
2.2. Configure as cache backend
To start using django-redis, you should change your Django cache settings to something like this:
CACHES = {
"default": {
"BACKEND": "django_redis.cache.RedisCache",
"LOCATION": "redis://127.0.0.1:6379/1",
"OPTIONS": {
"CLIENT_CLASS": "django_redis.client.DefaultClient",
}
}
}
django-redis, since 3.8.0, it starts using redis-py native url notation for connection strings, that allows better interoperability and have a connection string in more "standard" way.
redis://[:password]@localhost:6379/0 rediss://[:password]@localhost:6379/0 unix://[:password]@/path/to/socket.sock?db=0
Three URL schemes are supported:
-
redis://
: creates a normal TCP socket connection -
rediss://
: creates a SSL wrapped TCP socket connection -
unix://
creates a Unix Domain Socket connection
There are several ways to specify a database number:
-
A
db
querystring option, e.g. redis://localhost?db=0 -
If using the redis:// scheme, the path argument of the url, e.g.
redis://localhost/0
Note
|
if you are coming fron django-redis < 3.8.x, you are probably using redis_cache. Since
django-redis 3.8.x, redis_cache module is deprecated in favor to django_redis . The
redis_cache module will be removed in 3.9.x versions.
|
2.3. Configure as session backend
Django can by default use any cache backend as session backend and you benefit from that by using django-redis as backend for session storage without installing any additional backends:
SESSION_ENGINE = "django.contrib.sessions.backends.cache"
SESSION_CACHE_ALIAS = "default"
3. Advanced usage
3.1. Pickle version
For almost all values, django-redis uses pickle to serialize objects.
The latest available version of pickle is used by default. If you want set a concrete version, you
can do it, using PICKLE_VERSION
option:
CACHES = {
"default": {
# ...
"OPTIONS": {
"PICKLE_VERSION": -1 # Use the latest protocol version
}
}
}
3.2. Socket timeout
Socket timeout can be set using SOCKET_TIMEOUT
and SOCKET_CONNECT_TIMEOUT
options:
CACHES = {
"default": {
# ...
"OPTIONS": {
"SOCKET_CONNECT_TIMEOUT": 5, # in seconds
"SOCKET_TIMEOUT": 5, # in seconds
}
}
}
SOCKET_CONNECT_TIMEOUT
is the timeout for the connection to be established and
SOCKET_TIMEOUT
is the timeout for read and write operations after the connection
is established.
3.3. Compression support
django_redis comes with compression support out of the box, but is deactivated by default. You
can activate it setting COMPRESS_MIN_LEN
option to any value great than 0
.
CACHES = {
"default": {
# ...
"OPTIONS": {
"COMPRESS_MIN_LEN": 10,
}
}
}
zlib is used as default compression format. You can change it providing two callables, one for compress and an other for uncompress.
Let see an example, of how make it work with lzma compression format:
import lzma
CACHES = {
"default": {
# ...
"OPTIONS": {
"COMPRESS_MIN_LEN": 10,
"COMPRESS_COMPRESSOR": lzma.compress,
"COMPRESS_DECOMPRESSOR": lzma.decompress,
"COMPRESS_DECOMPRESSOR_ERROR": lzma.LZMAError
}
}
}
3.4. Memcached exceptions behavior
In some situations, when Redis is only used for cache, you do not want exceptions when Redis is down. This is default behavior in the memcached backend and it can be emulated in django-redis.
For setup memcached like behaviour (ignore connection exceptions), you should
set IGNORE_EXCEPTIONS
settings on your cache configuration:
CACHES = {
"default": {
# ...
"OPTIONS": {
"IGNORE_EXCEPTIONS": True,
}
}
}
Also, you can apply the same settings to all configured caches, you can set the global flag in your settings:
DJANGO_REDIS_IGNORE_EXCEPTIONS = True
3.5. Infinite timeout
django-redis comes with infinite timeouts support out of the box. And it behaves in same way as django backend contract specifies:
-
timeout=0
expires the value inmediatelly. -
timeout=None
infinite timeout
cache.set("key", "value", timeout=None)
3.6. Get ttl (time-to-live) from key
With redis, you can access to ttl of any stored key, for it, django-redis exposes ttl
function.
It returns:
-
ttl value for any volatile key (any key that has expiration)
-
0 for expired and not existent keys
-
None for keys that does not have expiration
>>> from django.core.cache import cache
>>> cache.set("foo", "value", timeout=25)
>>> cache.ttl("foo")
25
>>> cache.ttl("not-existent")
0
3.7. Scan & Delete keys in bulk
django-redis comes with some additional methods that help with searching or deleting keys using glob patterns.
>>> from django.core.cache import cache
>>> cache.keys("foo_*")
["foo_1", "foo_2"]
A simple search like this will return all matched values. In databases with a large number of keys
this isn’t suitable method. Instead, you can use the iter_keys
function that works like the keys
function but uses redis>=2.8 server side cursors. Calling iter_keys
will return a generator that
you can then iterate over efficiently.
>>> from django.core.cache import cache
>>> cache.iter_keys("foo_*")
<generator object algo at 0x7ffa9c2713a8>
>>> next(cache.iter_keys("foo_*"))
"foo_1"
For deleting keys, you should use delete_pattern
which has the same glob pattern syntax
as the keys
function and returns the number of deleted keys.
>>> from django.core.cache import cache
>>> cache.delete_pattern("foo_*")
3.8. Redis native commands
django-redis has limited support for some Redis atomic operations, such as the commands SETNX
and INCR
.
You can use the SETNX
command through the backend set()
method with the nx
parameter:
>>> from django.core.cache import cache
>>> cache.set("key", "value1", nx=True)
True
>>> cache.set("key", "value2", nx=True)
False
>>> cache.get("key")
"value1"
Also, incr
and decr
methods uses redis atomic operations when value that contains a key is suitable
for it.
3.9. Raw client access
In some situations your application requires access to a raw Redis client to use some advanced
features that aren’t exposed by the Django cache interface. To avoid storing another setting for
creating a raw connection, django-redis exposes functions with which you can obtain a raw client
reusing the cache connection string: get_redis_connection(alias)
.
>>> from django_redis import get_redis_connection
>>> con = get_redis_connection("default")
>>> con
<redis.client.StrictRedis object at 0x2dc4510>
Warning
|
Not all pluggable clients support this feature. |
3.10. Connection pools
Behind the scenes, django-redis uses the underlying redis-py connection pool implementation, and exposes a simple way to configure it. Alternatively, you can directly customize a connection/connection pool creation for a backend.
The default redis-py behavior is to not close connections, recycling them when possible.
3.10.1. Configure default connection pool
The default connection pool is simple. You can only customize the maximum number of connections
in the pool, by setting CONNECTION_POOL_KWARGS
in the CACHES
setting:
CACHES = {
"default": {
"BACKEND": "django_redis.cache.RedisCache",
...
"OPTIONS": {
"CONNECTION_POOL_KWARGS": {"max_connections": 100}
}
}
}
You can verify how many connections the pool has opened with the following snippet:
from django.core.cache import get_cache
from django_redis import get_redis_connection
r = get_redis_connection("default") # Use the name you have defined for Redis in settings.CACHES
connection_pool = r.connection_pool
print("Created connections so far: %d" % connection_pool._created_connections)
3.10.2. Use your own connection pool subclass
Sometimes you want to use your own subclass of the connection pool. This is possible with
django-redis using the CONNECTION_POOL_CLASS
parameter in the backend options.
from redis.connection import ConnectionPool
class MyOwnPool(ConnectionPool):
# Just doing nothing, only for example purpose
pass
# Omitting all backend declaration boilerplate code.
"OPTIONS": {
"CONNECTION_POOL_CLASS": "myproj.mypool.MyOwnPool",
}
3.10.3. Customize connection factory
If none of the previous methods satisfies you, you can get in the middle of the django-redis connection factory process and customize or completely rewrite it.
By default, django-redis creates connections through the django_redis.pool.ConnectionFactory
class that is specified in the global Django setting DJANGO_REDIS_CONNECTION_FACTORY
.
ConnectionFactory
class# Note: Using Python 3 notation for code documentation ;)
class ConnectionFactory(object):
def get_connection_pool(self, params:dict):
# Given connection parameters in the `params` argument,
# return new connection pool.
# It should be overwritten if you want do something
# before/after creating the connection pool, or return your
# own connection pool.
pass
def get_connection(self, params:dict):
# Given connection parameters in the `params` argument,
# return a new connection.
# It should be overwritten if you want to do something
# before/after creating a new connection.
# The default implementation uses `get_connection_pool`
# to obtain a pool and create a new connection in the
# newly obtained pool.
pass
def get_or_create_connection_pool(self, params:dict):
# This is a high layer on top of `get_connection_pool` for
# implementing a cache of created connection pools.
# It should be overwritten if you want change the default
# behavior.
pass
def make_connection_params(self, url:str) -> dict:
# The responsibility of this method is to convert basic connection
# parameters and other settings to fully connection pool ready
# connection parameters.
pass
def connect(self, url:str):
# This is really a public API and entry point for this
# factory class. This encapsulates the main logic of creating
# the previously mentioned `params` using `make_connection_params`
# and creating a new connection using the `get_connection` method.
pass
3.11. Pluggable parsers
redis-py (the Python Redis client used by django-redis) comes with a pure Python Redis parser that works very well for most common task, but if you want some performance boost, you can use hiredis.
hiredis is a Redis client written in C and it has its own parser that can be used with django-redis.
"OPTIONS": {
"PARSER_CLASS": "redis.connection.HiredisParser",
}
3.12. Pluggable clients
django_redis is designed for to be very flexible and very configurable. For it, it exposes a pluggable backends that make easy extend the default behavior, and it comes with few ones out the box.
3.12.1. Default client
Almost all about the default client is explained, with one exception: the default client comes with master-slave support.
To connect to master-slave redis setup, you should change the LOCATION
to something like this:
"LOCATION": [
"redis://127.0.0.1:6379/1",
"redis://127.0.0.1:6378/1",
]
The first connection string represents a master server and the rest to slave servers.
Warning
|
Master-Slave setup is not heavily tested in production environments. |
3.12.2. Shard client
This pluggable client implements client-side sharding. It inherits almost all functionality from the default client. To use it, change your cache settings to something like this:
CACHES = {
"default": {
"BACKEND": "django_redis.cache.RedisCache",
"LOCATION": [
"redis://127.0.0.1:6379/1",
"redis://127.0.0.1:6379/2",
],
"OPTIONS": {
"CLIENT_CLASS": "django_redis.client.ShardClient",
}
}
}
Warning
|
Shard client is still experimental, so be careful when using it in production environments. |
3.12.3. Herd client
This pluggable client helps dealing with the thundering herd problem. You can read more about it on Wikipedia.
Like previous pluggable clients, it inherits all functionality from the default client, adding some additional methods for getting/setting keys.
CACHES = {
"default": {
"BACKEND": "django_redis.cache.RedisCache",
"LOCATION": "redis://127.0.0.1:6379/1",
"OPTIONS": {
"CLIENT_CLASS": "django_redis.client.HerdClient",
}
}
}
This client exposes additional settings:
-
CACHE_HERD_TIMEOUT
: Set default herd timeout. (Default value: 60s)
3.13. Pluggable serializer
The pluggable clients serialize data before sending it to the
server. By default, django_redis serialize the data using Python
pickle
. This is very flexible and can handle a large range of object
types.
To serialize using JSON instead, the serializer JSONSerializer
is
also available.
CACHES = {
"default": {
"BACKEND": "django_redis.cache.RedisCache",
"LOCATION": "redis://127.0.0.1:6379/1",
"OPTIONS": {
"CLIENT_CLASS": "django_redis.client.DefaultClient",
"SERIALIZER": "django_redis.serializers.json.JSONSerializer",
}
}
}
There’s also support for serialization using MsgPack http://msgpack.org/ (that requires the msgpack-python library):
CACHES = {
"default": {
"BACKEND": "django_redis.cache.RedisCache",
"LOCATION": "redis://127.0.0.1:6379/1",
"OPTIONS": {
"CLIENT_CLASS": "django_redis.client.DefaultClient",
"SERIALIZER": "django_redis.serializers.msgpack.MSGPackSerializer",
}
}
}
4. License
Copyright (c) 2011-2015 Andrey Antukh <niwi@niwi.be>
Copyright (c) 2011 Sean Bleier
All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
1. Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
2. Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
3. The name of the author may not be used to endorse or promote products
derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.