rkhachatryan commented on pull request #16685:
URL: https://github.com/apache/flink/pull/16685#issuecomment-892406955


   Thanks for the review @curcur ,
   1. Yes, I first ran the build with changelog enabled
   2. Good point. I'll add this change to the first (hotfix) commit - 709a6e4
   
   I also see that the `StreamCheckpointingITCase` failure is related. It 
didn't fail nor when I enabled the backend by default nor in the private branch:
   -- the default value isn't written as metadata and therefore not passed to 
delegated states on recovery
   => map function will fail with NPE 
   => this NPE is swallowed by job recovery
   => however, it introduces a small delay which is sometimes enough to cause a 
deadlock in UC recovery (adding 10ns sleep has the same effect)
   => buffer requests eventually time out, but this is again swallowed by job 
recovery
   => eventually, `StreamCheckpointingITCase` times out (with unlimited 
recoveries)
   
   I'm going to fix this by serializing default value along with metadata in a 
separate hotfix commit.
   


-- 
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