Comments are misleading as memcache_servers option configures only the token backend, but not the cache backend
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
puppet-keystone |
Fix Released
|
Medium
|
Alex Schultz |
Bug Description
Currently 'memcache_servers' option doesn't configure 'memcache_servers' in [cache] section as one would assume at first glance, but it sets 'servers' in [keystone] section. I consider this as a bug as puppet documentation in keystone.pp is misleading and one can understand that the option is related to caching.
The comment regarding the 'memcache_servers' states that it used with a token driver,
#
# [*memcache_
# (optional) List of memcache servers in format of server:port.
# Used with token_driver 'keystone.
# Defaults to false. Example: ['localhost:11211']
#
but I misunderstood from the following comment that when I configure 'memcache_servers' and 'cache_backend' I will get a working cache
#
# [*cache_backend*]
# (optional) Dogpile.cache backend module. It is recommended that Memcache with pooling
# (keystone.
# This has no effects unless 'memcache_servers' is set.
# Defaults to 'keystone.
I think that we should
a) configure memcache_servers option to set the 'cache/
b) relax the dependency on memcache_servers from 'cache*' options and clearly state that these are not related,
b) or add an another option to configure caching servers and change the 'cache*' parameters to depend on the new option.
Changed in puppet-keystone: | |
status: | New → Triaged |
importance: | Undecided → Low |
importance: | Low → Medium |
Fix proposed to branch: master /review. openstack. org/262580
Review: https:/