Details
-
Improvement
-
Status: Closed
-
Major
-
Resolution: Fixed
-
1.11.1
Description
If the checkpoint directory from state.checkpoints.dir is not writable by the user that Flink is running with, checkpoints will be declined, but the real cause is not mentioned anywhere:
- the Web UI says: "Cause: The job has failed" (the Flink job is running though)
- the JM log says:
2020-08-14 12:13:18,820 INFO org.apache.flink.runtime.checkpoint.CheckpointCoordinator [] - Triggering checkpoint 2 (type=CHECKPOINT) @ 1597399998819 for job 2c567b14e8d0833404931ef47dfec266. 2020-08-14 12:13:18,921 INFO org.apache.flink.runtime.checkpoint.CheckpointCoordinator [] - Decline checkpoint 2 by task 0d4fd75374ad16c8d963679e3c2171ec of job 2c567b14e8d0833404931ef47dfec266 at a184deea621e3923fbfcb1d899348448 @ Nico-PC.lan (dataPort=35531).
- the TM log says:
2020-08-14 12:13:14,102 INFO org.apache.flink.streaming.runtime.tasks.SubtaskCheckpointCoordinatorImpl [] - Checkpoint 1 has been notified as aborted, would not trigger any checkpoint.
And that's it. It should have a real error message indicating that the checkpoint (sub)-directory could not be created.
Attachments
Issue Links
- is related to
-
FLINK-16753 Exception from AsyncCheckpointRunnable should be wrapped in CheckpointException
- Closed
- links to