mlowery commented on PR #1175:
URL: 
https://github.com/apache/flink-kubernetes-operator/pull/1175#issuecomment-6957947518

   Hi @Dennis-Mircea, could you take another look when you have a chance? The 
revision from Aug 13 looks like it addresses the kind collision, the callbacks 
after cleanup and concurrency from your review.
   
   To add to the motivation you asked about, we hit this in production 
independently. On a cluster with periodic savepoints, every new 
`FlinkStateSnapshot` adds a new set of JOSDK per-resource series, and history 
pruning deletes the CR but never the series. The series count only grows and 
resets only when the operator restarts. It eventually took the leader's 1 GiB 
heap: the OOM killed the JOSDK reconcile timer, and periodic reconciliation 
stopped for about 2.7 hours (no autoscaling, no periodic savepoints, no 
operator-driven recovery).
   
   Thanks @litiliu for driving this!


-- 
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