Skip to content

Core, Spark: Fix incorrect partial commit failure count in rewrite data files#16970

Open
john801205 wants to merge 2 commits into
apache:mainfrom
john801205:partial_commit_failure
Open

Core, Spark: Fix incorrect partial commit failure count in rewrite data files#16970
john801205 wants to merge 2 commits into
apache:mainfrom
john801205:partial_commit_failure

Conversation

@john801205

@john801205 john801205 commented Jun 26, 2026

Copy link
Copy Markdown

Summary

The rewrite data files procedure reported partial commit failures even when all commits succeeded, because it derived the total commit count as min(totalGroupCount, maxCommits). When groups are batched into groupsPerCommit = ceil(totalGroupCount / maxCommits) per commit, the actual number of commits is ceil(totalGroupCount / groupsPerCommit), which can be smaller. The mismatch made totalCommits - succeededCommits positive even with zero failures.

Compute totalCommits as ceil(totalGroupCount / groupsPerCommit) so the failed commit count is correct. Also make succeededCommits an AtomicInteger since it is incremented from multiple rewrite threads.

Issues

…ta files

The rewrite data files procedure reported partial commit failures even
when all commits succeeded, because it derived the total commit count as
min(totalGroupCount, maxCommits). When groups are batched into
groupsPerCommit = ceil(totalGroupCount / maxCommits) per commit, the
actual number of commits is ceil(totalGroupCount / groupsPerCommit),
which can be smaller. The mismatch made totalCommits - succeededCommits
positive even with zero failures.

Compute totalCommits as ceil(totalGroupCount / groupsPerCommit) so the
failed commit count is correct. Also make succeededCommits an
AtomicInteger since it is incremented from multiple rewrite threads.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
@john801205 john801205 force-pushed the partial_commit_failure branch 2 times, most recently from 268928a to beac357 Compare June 26, 2026 12:20
@john801205 john801205 force-pushed the partial_commit_failure branch from beac357 to c4c6538 Compare June 26, 2026 12:26
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant