Bug #45714
closed
[rbd-mirror] ensure ops are canceled during replayer shut down before waiting
Added by Jason Dillaman almost 4 years ago.
Updated almost 4 years ago.
Description
If rbd-mirror attempts to shut down its journal replayer, it might shut down after a op has started but before the OpFinish event is processed. In that case, the op will never finish and the wait-for-events state will never advance. This reverts to the pre-Octopus behavior prior to rbd-mirror being refactored.
- Status changed from New to In Progress
- Assignee set to Jason Dillaman
- Subject changed from Snapshot quiesce / unquiesce might stall librbd client shutdown to [rbd-mirror] ensure ops are canceled during replayer shut down before waiting
- Description updated (diff)
- Backport set to octopus
- Status changed from In Progress to Fix Under Review
- Pull request ID set to 35348
Note: backport to octopus should skip the first commit (the revert) but instead cherry-pick 4a23b4be68d00ac45216f7208ef19731006887da.
- Status changed from Fix Under Review to Pending Backport
- Copied to Backport #45885: octopus: [rbd-mirror] ensure ops are canceled during replayer shut down before waiting added
- Status changed from Pending Backport to Resolved
While running with --resolve-parent, the script "backport-create-issue" noticed that all backports of this issue are in status "Resolved" or "Rejected".
Also available in: Atom
PDF