Feature #61863
open
mds: issue a health warning with estimated time to complete replay
Added by Patrick Donnelly 11 months ago.
Updated 11 days ago.
Category:
Administration/Usability
Description
When the MDS is in up:replay, it does not give any indication to the operator when it will complete. We do have this information though. The MDS knows the end of the journal, the write position, and how quickly it's consuming (replaying) events. Issue a periodic health warning with the current completion percentage (e.g. 40% of journal read), time spent in replay, and expected time remaining.
Related issues
1 (1 open — 0 closed)
- Assignee set to Manish Yathnalli
- Status changed from New to In Progress
Patrick, the "MDS behind trimming" warning during up:replay is kind of expected in cases where there are lot many journal events/segments to replay. I think it makes sense to have this warning show up when the MDS is up:active and during up:replay the MDS could warn for longer replay times depending on the expected replay completion time/percentage.
Venky Shankar wrote:
Patrick, the "MDS behind trimming" warning during up:replay is kind of expected in cases where there are lot many journal events/segments to replay.
Sorry I'm not understanding where "MDS behind trimming" warning fits into this particular issue (beyond causing longer replay).
I think it makes sense to have this warning show up when the MDS is up:active and during up:replay the MDS could warn for longer replay times depending on the expected replay completion time/percentage.
If "this warning" meaning "MDS behind on trimming", I think we're on the same page.
- Pull request ID set to 52527
- Status changed from In Progress to Fix Under Review
- Assignee changed from Manish Yathnalli to Venky Shankar
- Backport changed from reef,quincy,pacific to reef,quincy
- Pull request ID changed from 52527 to 55616
Manish, I'm taking the ownership of this one. I retained your contribution tags in the new pull request (of course!).
- Related to Feature #65637: mds: continue sending heartbeats during recovery when MDS journal is large added
- Target version changed from v19.0.0 to v20.0.0
Also available in: Atom
PDF