[FLINK-23884] Fix the concurrent problem between triggering savepoint with drain and finish

• Type: Sub-task
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.14.0
• Fix Version/s: 1.14.0
• Component/s: Runtime / Checkpointing
• Labels: pull-request-available
• Assignee: Yun Gao
• Reporter: Yun Gao
• Created: 20/Aug/21 05:08
• Updated: 01/Sep/21 10:06
• Resolved: 27/Aug/21 02:44

Description

Currently when triggering stop-with-savepoint --drain, we would like to first finish the task and then trigger a savepoint. However, currently the implement does not ensures triggering get successful before the task get finished and close the mailbox, which might cause the savepoint fail.

Issue Links

• fixes: FLINK-23829 SavepointITCase JVM crash on azure
• links to: GitHub Pull Request #16905