Naman-555 opened a new pull request, #18909: URL: https://github.com/apache/iceberg/pull/18909
When `stopWorker()` throws during `CommitterImpl.close()`, for example because closing the writers fails on an S3 error, the coordinator is never stopped. Its thread keeps running with the closed task's catalog until the Connect worker restarts (#16016). `close()` now stops the coordinator and resets offsets even when stopping the worker fails, rethrowing the original exception, and `stopWorker()` clears the worker if `stop()` throws. #18039 fences such a coordinator once a new one starts; this stops it from being left behind. #18687 touches the leader check in the same method; the two are independent. Fixes #16016 Test plan: a coordinator must be terminated by `close()` when the worker's `stop()` throws, and the worker's exception must propagate. --- **AI Disclosure** - Model: Claude Opus 5.5 - Platform/Tool: Claude Code - Human Oversight: [unknown - human to fill in] - Prompt Summary: Diagnose a zombie coordinator seen in production and fix CommitterImpl.close() so the coordinator is stopped when stopping the worker fails, with a unit test. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
