DanielLeens commented on PR #12623: URL: https://github.com/apache/seatunnel/pull/12623#issuecomment-6607564961
Thank you @Vivek1106-04, this closes the evidence gap I was honest about last time. Your revert run is exactly the counterfactual I had only argued from source: with `PhysicalVertex.java` reset to the `dev` version and the test file unchanged, `testForceStopCompletesTheFutureOfATaskWhoseStateProcessIsStopped` and `testForceStopCompletesTheFutureOfATaskResetForRestore` time out, and with the PR's version all 20 pass. I checked the head `caa5f098fc0a5c6dd887ba8632dc43d463eb4580`: the two failing lines (`StateTransitionCleanupTest.java:189` and `:220`) are the `taskFuture.get(10, TimeUnit.SECONDS)` calls right after `physicalVertex.forceStop()`, so the timeout is precisely "the future of a stopped/reset vertex is never completed", which is the bug this PR fixes. The skipped-transition case staying green is also what the diff implies, since that path is unchanged. The merge-order line at the top of the description (#12511 must land first) is what I asked for; #12511 is still open, so F1 should be re-checked after it merges. My approval on this unchanged head stands; nothing else is needed from you right now, and please do not push the revert. Final approval and merge remain with a write-capable maintainer. -- 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]
