Description
The feature is described in the doc .
To test the feature, we should run a Flink program on a bounded source and configure it with a large checkpoint interval. With this feature, the program should end up immediately when no more records need to be processed, instead of waiting for one more periodic checkpoint.
Attachments
Attachments
Issue Links
- is caused by
-
FLINK-28386 Trigger an immediate checkpoint after all tasks has reached end-of-data
- Closed
- Testing discovered
-
FLINK-32945 NullPointerException when executing TopSpeedWindowing example with checkpointing enabled
- Closed