tillrohrmann commented on issue #8254: [FLINK-12219][runtime] Yarn application 
can't stop when flink job failed in per-job yarn cluste mode
URL: https://github.com/apache/flink/pull/8254#issuecomment-487940319
 
 
   As I've said before, the archiving handlers belong to Flink and if they are 
broken, then the system does not give any guarantees. If they were user 
provided, then I would agree that Flink should not fail in case of exceptions 
because they are not under its control.
   
   The problem is that we don't see this exception. The reason seems to be that 
we execute the `jobReachedGloballyTerminalState` in a `whenCompleteAsync` 
callback which returns a future containing the exception. Since we don't check 
this future, we will never see it. That's also why the 
`UncaughtExceptionHandler` is never called.
   
   I would suggest to add a utility `FutureUtils#assertNoException` which 
checks returned futures for normal completion. Please take a look at this 
commit 
https://github.com/tillrohrmann/flink/commit/524ead44f68ff02d52e8ff561c9e235e5b353123
 and let me know whether you would accept this as a fix for the problem.

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


With regards,
Apache Git Services

Reply via email to