fhan688 opened a new pull request, #3509:
URL: https://github.com/apache/fluss/pull/3509
### Purpose
Linked issue: https://github.com/apache/fluss/issues/3282
This PR completes the Hudi lake tiering write path by introducing the Hudi
`LakeCommitter`.
Together with the existing Hudi source, split planner/reader, writer, and
write result serializer changes, this enables Fluss tiering service to write
records into Hudi and commit Hudi instants as readable lake snapshots.
### Brief change log
- Add `HudiCommittable` and `HudiCommittableSerializer` for transferring
Hudi write statuses from lake writers to the committer.
- Add `HudiLakeCommitter` to:
- aggregate `HudiWriteResult`s into a committable;
- commit one Hudi instant with Fluss snapshot metadata;
- mark committed/aborted instants in `CkpMetadata`;
- discover missing Hudi snapshots committed by Fluss tiering.
- Wire Hudi committer creation through `HudiLakeTieringFactory`.
- Enable `HudiLakeStorage#createLakeTieringFactory`.
- Add the required `hadoop-mapreduce-client-core` provided dependency for
Hudi Parquet writing.
- Add tests for committable serialization, lake storage factory wiring,
and a local Hudi write-commit-recovery flow.
### Tests
- `mvn -pl fluss-lake/fluss-lake-hudi -am -DskipITs -Dcheckstyle.skip=true
-DfailIfNoTests=false test`
### API and Format
- Adds a Hudi-specific lake tiering committable type and serializer
version `1`.
- Enables Hudi `LakeTieringFactory` creation from `HudiLakeStorage`.
- No Fluss table storage format change.
- Hudi commits produced by Fluss tiering include extra metadata such as
`commit-user=__fluss_lake_tiering` and Fluss lake snapshot properties.
### Documentation
No documentation changes in this PR.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
