echauchot commented on PR #21037:
URL: https://github.com/apache/flink/pull/21037#issuecomment-1282298514

   > > What is the exception that you? Is it the assertion error about the 
final job state? Why not just allow both canceled & finished?
   > 
   > While testing an incorrect bounded source that does not report the 
expected metric (source under dev), running this test without disabling the 
killJob, here is what happens:
   > 
   > 1. the timeout fires
   > 2. the finally bloc runs the `killJob()` method that runs 
`CommonTestUtils.terminateJob() ` that triggers the job cancelation
   > 3. `Dispatcher#cancelJob()` throws 
`org.apache.flink.runtime.messages.FlinkJobTerminatedWithoutCancellationException:
 Flink job (8e3ca8038550e7e5aa3018b887ebf9a2) was not canceled, but instead 
FINISHED.`
   > 
   > So my reasoning was: on a bounded source, there is no point in killing the 
job as it will eventually finish. This will avoid the problem above. So when 
testing this incorrect source now we get 
j`ava.util.concurrent.TimeoutException: Timeout while comparing source metrics 
with 80 expected value` which is precisely the root cause of the problem in the 
source and not the `FlinkJobTerminatedWithoutCancellationException`
   
   @zentol WDYT ?


-- 
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