Hi team, could I get some reviews for this PR? <gi...
# contributing-to-starrocks
j
Hi team, could I get some reviews for this PR? github.com/StarRocks/starrocks/pull/71874 This fixes an issue where TABLET SAMPLE options are silently dropped for lake tables in shared-data clusters. I found this while repeatedly loading large tables during shared-data testing. Auto ANALYZE SAMPLE was actually doing a full tablet scan, which used enough resources that the following load could fail. In my test, ANALYZE SAMPLE went from around 2 minutes to 8 seconds after the sample options were correctly propagated. The change completes the FE -> BE propagation path for lake scans, matching the existing shared-nothing path. I also rebased it onto the latest main and added small FE and BE regression tests. This is a small fix, but I think it is important for real shared-data workloads. Would appreciate a review. Thanks!
jabgood 1