gharris1727 commented on PR #16788:
URL: https://github.com/apache/kafka/pull/16788#issuecomment-2272199968

   > requests for zombie fencing that any other workers send to it will fail.
   
   Okay, thanks for clarifying, that makes sense. I see that when the key is 
null, requests are rejected with a 503 error. I think that's still possible, if 
we're the first leader and don't get a chance to write a key to the topic 
before executing a request from a caller.
   
   Do you think that we could eliminate this error completely? If the key is 
null, we could reject it with a 403 and not reveal that the cluster is still 
starting up, triggering the retry logic on the requesting worker.
   
   We could also move the key verification to the herder thread, so that 
requests made during startup are delayed until the herder has finished 
starting, and we're sure that we've tried to install a key if one does not 
already exist. This might be less secure though, as it would allow a 
non-authenticated caller to saturate the herder tick thread.
   


-- 
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