Actions
Bug #13468
closedosd/ReplicatedPG.cc: 9128: FAILED assert(info.last_complete == info.last_update)
% Done:
0%
Source:
Community (dev)
Tags:
Backport:
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Pull request ID:
Crash signature (v1):
Crash signature (v2):
Description
This mainly happened when we marked some unfound objects lost, restarts helps.
I will update with more details after analyze the logs.
Updated by Guang Yang over 8 years ago
While I don't have verbose logs when the crash happened, I have the core dump, which shows:
(gdb) p pg_log.missing $2 = {missing = std::map with 0 elements, rmissing = std::map with 0 elements}
So we are pretty sure that the hobject_t passed to recover_got was not in the missing map (or else info.last_complete would be set to info.last_update). My guess is that it is removed from the pg_log.missing as part of the local filestore apply. That also explains why restart helps for this case.
Updated by Guang Yang over 8 years ago
A PR based on this - https://github.com/ceph/ceph/pull/6266
Updated by Nathan Cutler over 8 years ago
- Status changed from New to Fix Under Review
Updated by Guang Yang over 8 years ago
- Status changed from Fix Under Review to 7
Updated by Sage Weil about 7 years ago
- Status changed from 17 to Can't reproduce
this code was rewritten; assuming it's fixed.
Actions