[FLINK-26741] CheckpointIDCounter.shutdown should expose errors asynchronously

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.15.0
• Fix Version/s: 1.15.0
• Component/s: Runtime / Coordination
• Labels: pull-request-available
• Assignee: Matthias Pohl
• Reporter: Matthias Pohl
• Created: 19/Mar/22 15:24
• Updated: 24/Mar/22 09:08
• Resolved: 24/Mar/22 09:08

Description

CheckpointIDCounter.shutdown should return a CompletableFuture exposing errors if anything happens to enable retrying of the cleanup.

All implementations should also work in an idempotent fashion, i.e. nothing should happen if the artifact was already deleted.

This bug is not considered a 1.15 blocker because unit tests rerunning the shutdown worked properly already, i.e. in the worst case (of an error) we would just miss to clean up right now.

Issue Links

• blocks: FLINK-26114 DefaultScheduler fails fatally in case of an error when shutting down the checkpoint-related resources
• Testing discovered: FLINK-26798 JobMaster.testJobFailureWhenTaskExecutorHeartbeatTimeout failed due to missing Execution
• links to: GitHub Pull Request #19191
• links to: GitHub Pull Request #19196