DanielLeens commented on issue #11437: URL: https://github.com/apache/seatunnel/issues/11437#issuecomment-5137567639
Thanks for retesting this on the newer upstream `dev` build and for adding the 3-Master / 3-Worker evidence. I rechecked the linked fix state as well. PR #11458 is still open and draft, and its current Build check is not green, so this issue is still not fixed in `dev`. Your latest run is useful because it broadens the acceptance matrix beyond the earlier 3-Master / 2-Worker and high-occupancy multi-job cases. For the final verification after #11458 is updated and merged, I think we should now keep three concrete scenarios in scope: 1. the original 3-Master / 2-Worker single-job fixed-slot restore case; 2. the concurrent high-occupancy case where multiple jobs consume 15 of 16 slots; 3. this newer 3-Master / 3-Worker single-job case consuming 16 of 24 slots. The expected result is the same for all of them: after killing only the active Master, the restored job or jobs should leave `PENDING/RESTORE`, return to `RUNNING`, continue checkpoints, and release all fixed slots after cancel without requiring Worker restarts. Keeping this open until the fix lands and those restore/cleanup checks are verified on a build that actually contains the merged fix. -- 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]
