goutamadwant commented on issue #12010: URL: https://github.com/apache/seatunnel/issues/12010#issuecomment-5470914983
@DanielLeens @davidzollo I reproduced this on the current `dev` branch with both JDK 8 and JDK 11. During coordinator shutdown or a master transition, the submission executor can be interrupted immediately before `pendingJobQueue.put(...)`. `PeekBlockingQueue.put(...)` swallows the resulting `InterruptedException`, so the submission future completes successfully even though the local pending queue does not contain the job. The distributed job metadata remains in `PENDING`, so later recovery may still be possible. Before I start the fix, could you confirm the intended contract? 1. Fail the interrupted or stale submission with a retriable exception and perform ownership/epoch-safe cleanup. 2. Make acceptance non-interruptible, while fencing insertion and acknowledgement against coordinator shutdown or epoch changes. I can prepare the implementation and regression coverage once the expected behavior is confirmed. -- 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]
