Skip to content

Conversation

@xuang7
Copy link
Contributor

@xuang7 xuang7 commented Feb 9, 2026

What changes were proposed in this PR?

This PR fixes the issue where a failed LakeFS abort call could cause the database transaction to roll back, leaving the upload session stuck in the database. It moves the LakeFS abort call outside the database transaction in abortMultipartUpload to prevent orphaned upload sessions. The transaction now handles validation and DB cleanup, returning the necessary values (repoName, uploadId, physicalAddress) as a tuple. After the transaction commits, the LakeFS abort is called separately, so the session is always cleaned up regardless of whether LakeFS succeeds.

Any related issues, documentation, discussions?

Fixes #4196

How was this PR tested?

Manually tested

Was this PR authored or co-authored using generative AI tooling?

No

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.

Multipart upload abort can leave orphaned upload sessions when LakeFS abort fails

1 participant