rangareddy commented on issue #17208: URL: https://github.com/apache/hudi/issues/17208#issuecomment-5366107517
This issue was reviewed as part of the JIRA-migrated backlog triage (HUDI-9802). **Findings: needs a reproduction against a current build before it can be dispositioned.** The report is concrete: on a brand-new table version 9 table created with MDT off, turning MDT on and then running a DML with a WHERE filter fails with IllegalStateException: Partition record_index should be part of inflight metadata partitions here []. Reproducing it looks tractable - the repro is three steps with no data dependency - create a v9 table with MDT off, load data, enable MDT, run a filtered DML. I have not called this from the source. The surrounding code has been rewritten substantially since this was filed - table versions 8 and 9, the `BufferedRecordMerger` family replacing payload-based merging, the file-group-reader migration, the indexer abstraction in #18348 / #18372, and the `hudi-common` package moves in #19193 / #19195. Inferring either "still broken" or "silently fixed" through that much churn would be a guess. There is also no commit anywhere in the tree referencing HUDI-9802, so if it was fixed it was incidental rather than deliberate - another reason to verify rather than assume. The exception text is precise about the inconsistency: the file system view expects `record_index` among the inflight metadata partitions and finds an empty list. That points at MDT bootstrap-on-write not registering the partition as inflight before the tagging path consults it. Related: #17169 (HUDI-9740) is the mirror-image problem - a rolled-back index leaving a stale inflight MDT partition in the table config - so both are about that inflight set being out of step with reality. Keeping this open. -- 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]
