goutamadwant commented on PR #12288: URL: https://github.com/apache/seatunnel/pull/12288#issuecomment-5841703806
On point 5 above (flakiness evidence), I ran the loop locally. The test was `CheckpointCoordinatorFailoverIT#testStreamJobFailsAfterCheckpointTriggerDispatchFailure`, with each PR's diff applied to `dev` at `deb16a3c3`. JDK 11 runs used `-XX:ActiveProcessorCount=2` on a loaded 10-core macOS host, one fresh JVM per run. | variant | JDK 11 | JDK 11 + 6 CPU burners | JDK 8 | |---|---|---|---| | `dev` | 3 / 40 | 7 / 30 | 0 / 8 | | this PR | 2 / 40 | 5 / 30 | 1 / 8 | | #12449 | 0 / 40 | 0 / 30 | 0 / 8 | Every failure with this PR is the same `Checkpoint notify complete failed` message as on `dev`, now at line 857. That fits window (a) described above. `latestCompletedCheckpoint` is set before `notifyCompleted` runs, so the removal can still race checkpoint 1's own notification. #12449 waits for `pendingCounter == 0` and removes the entry under the coordinator lock. That closed the race in every run I did. It also needs no production accessor. I'd suggest going with #12449 and closing this one. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
