yahoo-eng-team team mailing list archive
-
yahoo-eng-team team
-
Mailing list archive
-
Message #45350
[Bug 1537617] [NEW] caching of the catalog does not invalidate across processes
Public bug reported:
Due to the way dogpile.cache works, the region-wide invalidation used in
the catalog caching backend is broken, since this will only invalidate
the local in-memory region and is not shared across backends.
This will produce odd effects when adding/removing endpoints if caching is enabled such as the endpoints needing to wait a significant period of time
for the cache to expire on some workers while others will see the change immediately.
** Affects: keystone
Importance: High
Status: Triaged
** Tags: liberty-backport-potential
** Description changed:
Due to the way dogpile.cache works, the region-wide invalidation used in
the catalog caching backend is broken, since this will only invalidate
the local in-memory region and is not shared across backends.
- This will produce odd effects when adding/removing endpoints if caching
- is enabled.
+ This will produce odd effects when adding/removing endpoints if caching is enabled such as the endpoints needing to wait a significant period of time
+ for the cache to expire on some workers while others will see the change immediately.
** Changed in: keystone
Status: New => Triaged
** Changed in: keystone
Importance: Undecided => High
** Tags added: liberty-backport-potential
--
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to OpenStack Identity (keystone).
https://bugs.launchpad.net/bugs/1537617
Title:
caching of the catalog does not invalidate across processes
Status in OpenStack Identity (keystone):
Triaged
Bug description:
Due to the way dogpile.cache works, the region-wide invalidation used
in the catalog caching backend is broken, since this will only
invalidate the local in-memory region and is not shared across
backends.
This will produce odd effects when adding/removing endpoints if caching is enabled such as the endpoints needing to wait a significant period of time
for the cache to expire on some workers while others will see the change immediately.
To manage notifications about this bug go to:
https://bugs.launchpad.net/keystone/+bug/1537617/+subscriptions
Follow ups