nileshkumar3 opened a new pull request, #22951:
URL: https://github.com/apache/kafka/pull/22951

   With the state updater enabled, Streams could log a spurious WARN 
(previously ERROR) — "Unable to acquire lock while closing the state store for 
task " — during back-to-back rebalances, when a task's per-task state-directory 
lock is transiently held by another thread as the task is handed off. It's 
harmless (no data loss or lag) but pollutes alerting.
   
   StateManagerUtil.closeStateManager() now short-circuits when there is 
nothing to close: if the state manager has no registered stores and we're not 
wiping state, it returns early without acquiring the lock or logging. When 
stores are still registered but the lock can't be acquired, the WARN is still 
emitted, preserving signal for genuinely unexpected failures. The unclean-EOS 
wipe path is unchanged.
   
   Add package-private ProcessorStateManager.hasRegisteredStores().
   Guard the lock/close/log in closeStateManager() with !wipeStateStore && 
!stateMgr.hasRegisteredStores().
   Update StateManagerUtilTest and add 
shouldSkipLockAcquisitionWhenNoRegisteredStoresToClose.


-- 
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