[FLINK-11630] TaskExecutor does not wait for Task termination when terminating itself

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.8.0
• Fix Version/s: 1.9.1, 1.10.0
• Component/s: Runtime / Coordination
• Labels: pull-request-available
• Assignee: boshu Zheng
• Reporter: Till Rohrmann
• Created: 15/Feb/19 15:49
• Updated: 20/Nov/20 12:14
• Resolved: 18/Aug/19 14:21

Description

The TaskExecutor does not properly wait for the termination of Tasks when terminating. In fact, it does not even trigger the cancellation of the running Tasks. I think for better lifecycle management it is important that the TaskExecutor triggers the termination of all running Tasks and then wait until all Tasks have terminated before it terminates itself.

Issue Links

• causes: FLINK-11631 TaskExecutorITCase#testJobReExecutionAfterTaskExecutorTermination unstable on Travis
• causes: FLINK-12285 Memory leak in SavepointITCase and SavepointMigrationTestBase
• duplicates: FLINK-12707 Close minicluster will cause memory leak when there are StreamTask closed abnormal
• is duplicated by: FLINK-12332 Cancel running tasks if exist before shutting down TM
• relates to: FLINK-12707 Close minicluster will cause memory leak when there are StreamTask closed abnormal
• links to: GitHub Pull Request #7757
• links to: GitHub Pull Request #9072