wombatu-kun commented on PR #18849: URL: https://github.com/apache/iceberg/pull/18849#issuecomment-6788894815
This `finally` covers the failure behind the timeouts. On 2.2 and 2.3, `StreamTask.cleanUp` calls `disableInterruptOnCancel()` and `Thread.interrupted()` before `resourceCloser.close()`, where `closeAllOperators` is registered, so cancellation can't interrupt `awaitTermination` inside `close()`. The double close can only come from `processWatermark` leaving a closed service behind. Only 2.2 and 2.3 hang, although #18815 backported `testConvertEqualityDeletesWithRewriteEnabledE2e` to 1.20 as well. Since 2.2, `Task.doRun` closes `invokable.cleanUp` inside its failure handler, so an exception from `close()` ends in "FATAL - exception in exception handler of task" and stops the only TaskManager, while 1.20 adds it as a suppressed exception. The 2.2 tree has hit the same timeout on main since Nov 9, so it needs this fix too. -- 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]
