Skip to content

Conversation

@tub
Copy link
Contributor

@tub tub commented Jan 30, 2026

Purpose

The latest hadoop s3a libraries expose conditional writes https://issues.apache.org/jira/browse/HADOOP-19256.
We can utilise this in order to avoid potential data loss with snapshots being overwritten.

Tests

Added coverage

API and Format

No changes to the physical storage or APIs

Documentation

TODO - update docs

tub and others added 6 commits January 30, 2026 17:49
- Add FileIOTest.testConditionalWriteDefaults for tryToWriteAtomicIfAbsent
- Add RenamingSnapshotCommitTest for conditional vs lock-based commit paths

Co-Authored-By: Claude <[email protected]>
Uses same Hadoop 3.4+ createFile().overwrite(false) API as S3.

Co-Authored-By: Claude <[email protected]>
Hadoop 3.4.x switched from AWS SDK v1 to v2, requiring updates to:
- S3MultiPartUpload: Use SDK v2 types and new WriteOperationHelper API
- S3TwoPhaseOutputStream: Use CompleteMultipartUploadResponse
- S3MultiPartUploadCommitter: Use CompletedPart instead of PartETag

Co-Authored-By: Claude <[email protected]>
- Update Hadoop from 3.3.4 to 3.4.2
- Update hadoop-shaded-guava from 1.1.1 to 1.4.0
- Update hadoop-shaded-protobuf from 3_7 to 3_25
- Add software.amazon.awssdk:bundle:2.29.52

Co-Authored-By: Claude <[email protected]>
- Update Hadoop from 3.3.4 to 3.4.2
- Update hadoop-shaded-guava from 1.1.1 to 1.4.0
- Update hadoop-shaded-protobuf from 3_7 to 3_25

Co-Authored-By: Claude <[email protected]>
@tub tub closed this Feb 2, 2026
@tub
Copy link
Contributor Author

tub commented Feb 2, 2026

Closing in favour of #7187

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.

1 participant