[FLINK-31995] DirectExecutorService doesn't follow the ExecutorService contract throwing a RejectedExecutionException in case it's already shut down

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.17.0, 1.16.1, 1.18.0
• Fix Version/s: 1.18.0
• Component/s: Runtime / Coordination
• Labels: pull-request-available
• Assignee: Matthias Pohl
• Reporter: Matthias Pohl
• Created: 04/May/23 07:45
• Updated: 30/May/23 13:31
• Resolved: 04/May/23 15:31

Description

We experienced an issue where we tested behavior using the DirectExecutorService with the ExecutorService being shutdown already. The tests succeeded. The production code failed, though, because we used a singleThreadExecutor for which the task execution failed after the executor was stopped. We should add this behavior to the DirectExecutorService as well to make the unit tests be closer to the production code.

Issue Links

• causes: FLINK-32204 ZooKeeperLeaderElectionTest.testZooKeeperReelectionWithReplacement fails with The ExecutorService is shut down already. No Callables can be executed on AZP
• Discovered while testing: FLINK-31838 Move thread handling from DefaultMultipleComponentLeaderElectionService into DefaultLeaderElectionService
• relates to: FLINK-29813 Remove deprecated Curator API usages
• links to: GitHub Pull Request #22517