KAFKA-20735: Skip state-directory lock on close when there are no sto… - #22951
Open
nileshkumar3 wants to merge 1 commit into
Open
KAFKA-20735: Skip state-directory lock on close when there are no sto…#22951nileshkumar3 wants to merge 1 commit into
nileshkumar3 wants to merge 1 commit into
Conversation
Contributor
|
@nileshkumar3 thank you for the PR. LGTM |
Nikita-Shupletsov
approved these changes
Jul 28, 2026
Contributor
Author
|
Thanks @Nikita-Shupletsov for the review. |
Contributor
Author
|
@lucasbru Could you please review when get a chance. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
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.
Reviewers: Nikita Shupletsov nikita@shupletsov.ca