-
Epic
-
Resolution: Unresolved
-
Undefined
-
None
-
None
-
None
-
None
-
Fix default cache config to work with memcached running over IPv6 and TLS disabled
-
19
-
False
-
-
False
-
Not Selected
-
Proposed
-
Proposed
-
To Do
-
Proposed
-
Proposed
-
0% To Do, 50% In Progress, 50% Done
-
-
Currently service operators are configured as below when running with tls disabled and ipv6 enabled:-
[cache] backend = dogpile.cache.memcached enabled=true memcache_servers=memcached-0.memcached.openstack.svc:11211,memcached-1.memcached.openstack.svc:11211,memcached-2.memcached.openstack.svc:11211 tls_enabled=false
But with this config services(for ex keystone/neutron) do not work or take too much time to respond and hit the api timeout.
When backend is "dogpile.cache.memcached" and memcache serving over IPv6 it requires memcache_servers to contain inet6: prefix as described in https://review.opendev.org/c/openstack/puppet-tripleo/+/833711/11/manifests/profile/base/keystone.pp
With TLS enabled, issue is not seen as "dogpile.cache.pymemcache" backend is configured and that do not require inet6: prefix.
It got missed when TLS support was enabled https://issues.redhat.com/browse/OSPRH-5283
This Epic is to track the fix of the issue.
- is triggered by
-
OSPRH-5283 service operators connection to memcached over TLS
- Closed
- links to
- mentioned on