yihua opened a new issue, #924: URL: https://github.com/apache/hudi-rs/issues/924
Write stats undercount deletes that come from `_hoodie_is_deleted` delete markers in two cases: - On copy-on-write, a delete marker whose partition differs from the stored row's removes the stored row, but `WriteResult.num_deletes` stays 0 and the old file group's write stat records it as an update. A CDC delete that carries only the key hits this, since its null partition value routes to `__HIVE_DEFAULT_PARTITION__`. - Merge-on-read reports `num_deletes: 0` for every delete marker. Java counts both in `numDeletes`. Both `WriteResult` and the commit's write stats should count a delete wherever a marker removes a row. Found while reviewing #851, which fixes the in-place copy-on-write case. -- 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]
