zhuxiangyi commented on PR #9667: URL: https://github.com/apache/paimon/pull/9667#issuecomment-5749051768
Agreed, and the diagnosis is exact: the switch exists because the direct postpone path recomputes the bound per attempt, where Flink's `PostponeBucketSink` never sets one. The three commit switches this PR adds are all patches over the same mismatch — a batch committer closed after every micro-batch driving a filtered commit; a sink that keeps one `StreamTableCommit` across batches would drop them, let maintenance run asynchronously again, and, writing to bucket -2 as Flink does, also cover the staged path, which this PR only warns on for an append into a postpone table with existing buckets. Opened #10010 with the scope you listed — complete-mode overwrite semantics, callback retries, maintenance, postpone path performance — and I'll take it once this lands. -- 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]
