Details
-
Bug
-
Status: Closed
-
Major
-
Resolution: Duplicate
-
None
Description
When refactoring the checkpoint triggering logic in FLINK-21067, it tries to also support the FINISHED state, but it change the abort condition from == RUNNING to != (CREATED, SCHEDULED, DEPLOYING), which makes the checkpoint could not be aborted in these cases. This might happen if during trigger there is also job failover. It might cause trigger failure and make checkpoint has to wait for timeout.
Attachments
Issue Links
- duplicates
-
FLINK-22003 UnalignedCheckpointITCase fail
- Resolved
- relates to
-
FLINK-22003 UnalignedCheckpointITCase fail
- Resolved
- links to