SEZ9 commented on issue #12164: URL: https://github.com/apache/seatunnel/issues/12164#issuecomment-5578014994
Thanks for the detailed analysis, @DanielLeens. The gap you describe is plausible: the `catch (Throwable t)` only runs cleanup when `!contextPublished`, so an exception from `onContextPublished`, `submitThreadShareTask`, or `submitBlockingTask` (e.g. a `RejectedExecutionException` from an executor shutting down during failover) would leave stale `executionContexts`/`cancellationFutures` entries and make every later redeploy for that `TaskGroupLocation` hit the skip branch. A rollback that removes both map entries on the post-publish failure path sounds like the right fix. @Rangsh, thanks for volunteering to pick this up. When you open a PR, it would be great to include a test that injects a throwing executor after context publication and asserts that a subsequent `deployTask` for the same `TaskGroupLocation` actually redeploys, so the fix is covered against regressions. <!-- streview-comment:876 --> -- 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]
