fresh-borzoni opened a new pull request, #3317: URL: https://github.com/apache/fluss/pull/3317
## Summary When a lake-enabled primary-key table has no lake snapshot yet (e.g. not yet tiered), batch reads fell back to scanning the entire log from EARLIEST for every bucket, ignoring any Fluss kv snapshots that already existed. For tables with kv snapshots taken but no lake snapshot, this re-read all historical changes. Plumb the same per-bucket dispatch that FlussUpsertBatch already uses into FlussLakeUpsertBatch.planFallbackPartitions: hybrid (kv snapshot + log tail bounded by stoppingOffset) where a snapshot exists, log-only from EARLIEST bounded by stoppingOffset otherwise. The reader side (FlussUpsertPartitionReader) already supports both shapes. -- 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]
