rahil-c opened a new pull request, #20192: URL: https://github.com/apache/hudi/pull/20192
### Describe the issue this Pull Request addresses closes #20191 `read_blob()` on OUT_OF_LINE blobs fails under `KryoSerializer` when Hudi's Kryo registrator is not configured. The broadcast `HadoopStorageConfiguration` keeps its Hadoop `Configuration` in a `transient` field that only Java serialization restores, so executors get a null configuration. ### Summary and Changelog `read_blob()` now works under any `spark.serializer`. - `BatchedBlobReader`: adds `broadcastStorageConf`, which broadcasts the Hadoop configuration wrapped in Spark's `SerializableConfiguration`, and `storageConfFrom`, which rebuilds the `HadoopStorageConfiguration` on the executor. This matches how the rest of Hudi's read path ships configuration (for example `HoodieFileGroupReaderBasedFileFormat`). - `BatchedBlobReader.readBatched`, `processRDD` and `BatchedBlobReadExec.doExecute` use the new broadcast. - `TestBatchedBlobReaderKryo`: reads OUT_OF_LINE blobs after a Kryo round trip of the broadcast value with no Hudi registrator. Local mode does not serialize broadcasts, so the test does the round trip itself. The new test fails on master with a `NullPointerException` from the null configuration (`HadoopStorageConfiguration.writeObject`) and passes with this change. `TestBatchedBlobReaderKryo`, `TestReadBlobSQL` and `TestBatchedBlobReader` pass (37 tests). ### Impact None on public APIs. `BatchedBlobReader.processRDD` now takes `Broadcast[SerializableConfiguration]`; it is an internal helper. ### Risk Level low ### Documentation Update none ### Contributor's checklist - [x] Read through [contributor's guide](https://hudi.apache.org/contribute/how-to-contribute) - [x] Enough context is provided in the sections above - [x] Adequate tests were added if applicable -- 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]
