[flink] Skip LogSplit for empty buckets in lake union read #2957
Description
Search before asking
- I searched in the issues and found nothing similar.
Description
LakeSplitGenerator.generateSplit() creates a LogSplit for buckets with no lake data even when stoppingOffset is 0 (empty bucket). This causes the reader to block for the full poll timeout (default 10s) per empty bucket before returning nothing. With many empty buckets, this adds significant latency to batch queries.
The non-null branch already has this guard, but null branch is missing it.
Willingness to contribute
- I'm willing to submit a PR!