[FLINK-17043] Putting more information into accounting when failing a job in FailureManager

• Type: Improvement
• Status: Open
• Priority: Not a Priority
• Resolution: Unresolved
• Affects Version/s: 1.13.0
• Component/s: Runtime / Checkpointing
• Labels: auto-deprioritized-major auto-deprioritized-minor
• Assignee: Unassigned
• Reporter: Congxian Qiu
• Created: 08/Apr/20 03:18
• Updated: 16/Nov/21 10:44

Description

Currently, we only fail the job when we received continues "CHECKPOINT_DECLINED" message, but ignored the "timeout"/"task_failure"/"task_checkpoint_failure"/"finalize_checkpoint_failure" and so on.

In my opinion, we should put some checkpoint failure reason above into account when failing a job (not only the "CHECKPOINT_DECLINED" reason"

This issue is inspired by a [user mail list|http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/Making-job-fail-on-Checkpoint-Expired-tt34051.html],

 

Issue Links

• is related to: FLINK-17351 CheckpointCoordinator and CheckpointFailureManager ignores checkpoint timeouts