DanielLeens commented on PR #11841: URL: https://github.com/apache/seatunnel/pull/11841#issuecomment-6406837368
Thanks @JeremyXin, that closes both of my open points on head `881e275f`. 1. **`handleSaveMode` guard** - Your confirmation that `JobMaster.init` is the only direct production caller of `handleSaveMode(..., RestoreMode)` matches what I traced, so keeping the non-restore guard is right. +1 to @SEZ9's small ask: please put the "restore branch is preserved public-helper behavior, not reached from the guarded init path" note in the Javadoc of the new overload, so the next reader does not assume `init` exercises it. 2. **`CheckpointCoordinator` rename** - Agreed, it is a pure parameter rename with the same value flowing through `JobMaster.initCheckPointManager` -> `CheckpointManager`. Adding your one-sentence call-chain explanation to the PR description is a nice touch, and +1 to @SEZ9's request to also mention in the description (and release notes if applicable) that a checkpoint-mode restore hitting a pending cleanup record or existing job metrics now fails with `JobException` instead of taking the savepoint path. CI note for this head: the fork `Build` run 40111365132 has exactly one failed job, `engine-v2-it (11, ubuntu-latest)`, with a single failing test, `BackpressureSlowSinkIT.testCheckpointsKeepCompletingUnderSustainedBackpressure` (observed 2 of the expected 3 additional checkpoints in the 90s window; the other 206 tests in that lane passed). This PR does not touch checkpoint triggering or barrier injection, so this looks like a known timing-sensitive flake rather than a diff-related failure (an inference from the diff and the assertion, not reproduced). The lowest-cost next step is to re-run only that failed job on the fork rather than the whole workflow. Once the description/Javadoc notes are in and that lane is green, I have nothing further from my side. -- 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]
