Redis is used for two distinct purposes and the different configurations shouldn’t be mixed.
One is used as an LRU cache for the Symbolication service. It’s basically an alternative to a disk cache where eviction is automatically taken care of. The LRU functionality is dependent on two things:
maxmemoryconfiguration being something other than 0 otherwise it will just continue to fill up until the server/computer runs out of RAM.
maxmemory-policysetting being someting other than
The other Redis server is used for miscellaneous caching and as a broker for message queue workers (TO BE UPDATED).
Predicted Production Use¶
In 2014, Ted calculated that we need approximately 35GB storage to have a 99% cache hit ratio of all symbols that Socorro needs when symbolicating. In Tecken we don’t store downloaded symbol files, but instead we store a minor subset of the downloaded files so by estimates we only store 20% of that weight. In conclusion, to maintain a 99% cache hit ratio we need 6GB and 2GB for a 95% cache hit ratio.
Usage In Django¶
Within the Django code, these two are accessible in this way:
from django.core.cache import caches regular_cache = caches['default'] lru_cache = caches['store']
The first (“default”) cache is also available in this form:
from django.core.cache import cache regular_cache = cache
Because it uses the django-redis and the Django Cache framework API
you can use it for all other sorts of caching such as caching sessions or
cache control for HTTP responses. Another feature of this is that you can
bypass the default expiration by explicitly setting a
which means it’s persistent. For example:
from django.core.cache import cache cache.set('key', 'value', timeout=None)
To go into CLI of each two Redis database use these shortcuts:
$ make redis-cache-cli (...or...) $ make redis-store-cli
From there you can see exactly what’s stored. For example, to see the list of all symbols stored in the LRU cache:
$ make redis-store-cli redis-store:6379> keys * 1) ":1:symbol:xul.pdb/44E4EC8C2F41492B9369D6B9A059577C2" 2) ":1:symbol:wntdll.pdb/D74F79EB1F8D4A45ABCD2F476CCABACC2"
The default configuration, in Docker, for the Redis service used as a LRU
cache is defined in the
docker/images/redis/redis.conf file and it
maxmemory number that is appropriate for local development.
When deployed in production this should be better tuned to fit the server
it’s on. This configuration also sets the right
allkeys-lru which is also ideal for production usage.
To see the configuration, use the
redis-store service in the shell:
$ make redis-store-cli redis-store:6379> config get maxmemory 1) "maxmemory" 2) "524288000" redis-store:6379> config get maxmemory-policy 1) "maxmemory-policy" 2) "allkeys-lru"
To override this, simply use
config set instead of
$ make redis-store-cli redis-store:6379> config set maxmemory 100mb OK redis-store:6379> config get maxmemory 1) "maxmemory" 2) "104857600"
To get an insight into the state of the Redis service use the
$ make redis-store-cli redis-store:6379> info # Server redis_version:3.2.8 redis_git_sha1:00000000 redis_git_dirty:0 redis_build_id:9c531c9c1d171a62 redis_mode:standalone os:Linux 4.9.13-moby x86_64 arch_bits:64 multiplexing_api:epoll <redacted>
If you stop the Docker service and start it again it will revert to the
Unit Testing in Docker¶
Since Redis is the actual cache backend used even in unit tests, its
data is persistent between tests. To avoid confusion between unit tests
clear_redis_store pytest fixture. For example:
from django.core.cache import cache def test_storage1(clear_redis_store): assert not cache.get('foo') cache.set('foo', 'bar') def test_storage2(clear_redis_store): assert not cache.get('foo') cache.set('foo', 'different')