Actions
Bug #14319
closedDouble decreased the count to trim caps which will cause failing to respond to cache pressure
Status:
Resolved
Priority:
Normal
Assignee:
-
Category:
-
Target version:
-
% Done:
0%
Source:
Community (dev)
Tags:
Backport:
hammer
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Component(FS):
Labels (FS):
Pull request ID:
Crash signature (v1):
Crash signature (v2):
Description
When reaching mds cache size, mds will ask clients to trim its own caps. In ceph-fuse, it will recalculate current caps size for each loop, so it double decreases the loop counts. Hence at most time, ceph-fuse client only trims half of expected caps number. MDS often reports the warning of failing to respond to cache pressure and have to wait for asking client to trim caps next time.
Updated by Samuel Just over 8 years ago
- Project changed from Ceph to CephFS
- Category deleted (
11)
Updated by Zhi Zhang over 8 years ago
- Backport set to hammer
This issue also exists on master, so close original PR and create this new one.
Updated by Greg Farnum about 8 years ago
- Status changed from New to Pending Backport
Whoops, this got merged in way back then in January!
Looks like no backport got scheduled or anything; we'll see if it makes it — there are plenty of other reasons to upgrade to Jewel when using CephFS!
Updated by Nathan Cutler about 8 years ago
- Copied to Backport #15512: hammer: Double decreased the count to trim caps which will cause failing to respond to cache pressure added
Updated by Loïc Dachary over 7 years ago
- Status changed from Pending Backport to Resolved
Actions