benWize commented on pull request #15664:
URL: https://github.com/apache/beam/pull/15664#issuecomment-938138029


   Hi, @ibzib would you help me to review this? 
   I think the cause is that `triggerSavepoint` in 
   
https://github.com/apache/beam/blob/master/runners/flink/src/test/java/org/apache/beam/runners/flink/FlinkSavepointTest.java#L259
    is called twice as shown in the logs here 
https://scans.gradle.com/s/tdagq66c7f4n2/tests/:runners:flink:1.13:test/org.apache.beam.runners.flink.FlinkSavepointTest/testSavepointRestoreLegacy/1/output
 (A log with the message: `Triggering cancel-with-savepoint` is shown twice). 
   The first loop tries to trigger the savepoint and cancel, if the first call 
in the loop throws an exception, a second loop tries to trigger the savepoint 
while some tasks are been canceled, which causes a fail with the message 
`Failed to trigger checkpoint for job xxx since Checkpoint triggering task 
Source: Impulse (1/1) of job xxx is not being executed at the moment. Aborting 
checkpoint. Failure reason: Not all required tasks are currently running.`
   I was able to reproduce the error shown in the logs locally, executing 
`triggerSavepoint` twice. 
   My proposed fix is to split trigger and cancel to prevent canceling if the 
savepoint operator fails on its first call.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


Reply via email to