akalash commented on pull request #17440: URL: https://github.com/apache/flink/pull/17440#issuecomment-940200558
> Do we really just stop debloating or do we trigger a failover? As I can see we just stopping debloating and no failover would be triggered > Shouldn't the exception from the mailbox processor fail the task? I was also surprised but it looks like we just ignore this error(it is exactly what should be figured out later). > Why does it happen only for announcing the new buffer size? if we have input data already then the this NPE is impossible, it is exactly what happens when we call other methods except for the new buffer size. So everything except the announcement of the buffer size is a reaction to the input data. > Shouldn't we have the same issue for all other methods of CreditBasedPartitionRequestClientHandler such as e.g. notifyCreditAvailable? As you can see I made this NPE impossible by waiting for channel initialization on the start. So everything should work well now > Is it an error if we continue processing? IMO, at best it should be warn, I am event thinking if it shouldn't be just a debug. What are users supposed to do with the line in logs? I don't really sure that we need this try/catch block at all. My main fix is about waiting for the initialization but these changes I made just in case, but perhaps it is much better to understand the flink behavior when we have exceptions inside of the action.( But in general, I agree that `debug` suits here more) -- 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: issues-unsubscr...@flink.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org