keith-turner commented on issue #4076:
URL: https://github.com/apache/accumulo/issues/4076#issuecomment-1861239995

   > It's possible that there could be a race condition between terminating a 
pod and reducing the number of replicas in the deployment. I'm wondering if we 
might need an RPC that tells the server to not accept any new work (shutting 
down its Thrift server?).
   
   I think we could defer making decision on an RPC to handle that case.  Race 
conditions will always be there when making decisions based on metrics that 
lag.  A goal could be to choose pods to shutdown in a way that is better than 
randomly choosing on average.    There may still be false positives, just want 
less false positives than choosing randomly.  Whatever end to end strategy is 
used could try to determine how much it improves over randomly choosing as a 
way of evaluating how good it is.


-- 
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