SEZ9 commented on PR #12152:
URL: https://github.com/apache/seatunnel/pull/12152#issuecomment-5852092838

   Thanks for the follow-up on `0a4b61553` — I re-read the current head against 
the two points I raised earlier, and both look resolved:
   
   - **F1 (rejected `execute(...)` silently dropping the report):** the offload 
is now guarded — a `RejectedExecutionException` from the primary pool falls 
through to the static daemon `ERROR_REPORT_FALLBACK_EXECUTOR`, and the future 
is completed after dispatch, so a `.join()`/`.get()` waiter can no longer hang 
when the shared pool has been `shutdownNow()`-ed on a master switch. 
Dispatch-before-complete ordering keeps the `isDone()` guard meaningful.
   - **F2 (test only proved the caller returned):** 
`testCheckpointErrorReportDoesNotRunOnCallerThread` now positively asserts 
delivery via `Mockito.verify(checkpointManager, 
Mockito.timeout(5000)).handleCheckpointError(eq(1), eq(false))` once the busy 
executor is released, and 
`testCheckpointErrorReportCompletesFutureWhenExecutorRejects` / 
`testCheckpointErrorReportRejectionStillReachesCancellationOffCallerThread` 
cover the shutdown/rejection path end to end. That is exactly the coverage I 
was after.
   
   One small, non-blocking note on the same topic: the double-rejection 
last-resort branch (`CheckpointCoordinator.java:585-604`) isn't exercised by 
any test and is unreachable today since nothing shuts down the fallback 
executor; if it is ever reached, the exception message recorded at line 597 
looks wrong. Happy for that to land as a fast follow-up rather than in this PR 
— just confirm you're fine tracking it separately.
   
   Remaining ask before I approve: the `Build` check on this head is still the 
earlier completed run (`35681392273`), whose 3 failures don't touch 
`CheckpointCoordinator.java` or its test. Since the branch has drifted behind 
`dev`, could you rebase and push so we get a fresh run? A green (or 
confirmed-unrelated) result on the rebased head and I'll approve.
   
   <!-- streview-comment:1352 -->


-- 
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]

Reply via email to