[GitHub] [kafka] dstelljes commented on pull request #12295: KAFKA-13586: Prevent exception thrown during connector update from crashing distributed herder

[GitBox <gi...@apache.org>]

dstelljes commented on PR #12295:
URL: https://github.com/apache/kafka/pull/12295#issuecomment-1160666379

   Hey @C0urante, thanks for the comments!
   
   I just added a variation on the update test for this, but it’s causing `testTaskRequestedZombieFencingForwardedToLeader` to hang locally. I’m still not sure what I messed up, but I’m pushing it up as is since the zombie fencing tests are new and you’ll probably be able to spot the problem much faster than I will.
   
   Ideally, I think `ConfigException` itself would be checked. That way, `ConfigProvider::get` implementations could throw `ConfigException` for invalid paths, missing keys, etc., and other runtime exceptions could be assumed to be a provider defect or some other worker-level issue. That ship has probably sailed though...
   
   It seems fair to wrap everything in `startConnector` and funnel every exception to the callback since `reconfigureConnector` does the same thing. In `getConnectorStartingCallable`, immediate failures are passed to `onFailure`; is that something we’d have to account for?


-- 
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: jira-unsubscribe@kafka.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org