Skip to content

Conversation

@KMGeon
Copy link
Contributor

@KMGeon KMGeon commented Dec 29, 2025

Change filterCount field from long to AtomicLong to ensure thread-safety when incrementing filter count in concurrent chunk processing.

Resolves #5188

@quaff
Copy link
Contributor

quaff commented Dec 29, 2025

What about other counts?

@KMGeon KMGeon force-pushed the GH-5188 branch 2 times, most recently from 358b8d8 to ab01952 Compare December 30, 2025 15:44
@KMGeon
Copy link
Contributor Author

KMGeon commented Dec 30, 2025

다른 수치는 어떻습니까?

Thank you for your feedback! I've added processSkipCount to AtomicLong as well in the latest commit.

Regarding other count fields (readSkipCount, writeSkipCount, writeCount), they are only modified by single-threaded operations in ChunkOrientedTasklet, so they don't have the same race condition issue that filterCount and processSkipCount have in concurrent chunk processing scenarios.

Let me know if you think those should be converted as well for consistency.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

StepContribution counters are not thread-safe during parallel chunk processing

2 participants