eomiks opened a new pull request, #8596: URL: https://github.com/apache/hbase/pull/8596
Fixes the ROWS_SCANNED server-side scan metric double-counting rows that pass the filter when the scan uses the joined heap, i.e. `Scan#setLoadColumnFamiliesOnDemand(true)` with a filter marking some families as non-essential (`Filter#isFamilyEssential`). `RegionScannerImpl#populateResult` increments the metric whenever the given heap finishes the current row, and `populateFromJoinedHeap` goes through the same method for the non-essential families, so a row passing the filter was counted twice (once for the store heap, once for the joined heap). The row-scanned event is now recorded only when populating from the store heap. The new test in `TestServerSideScanMetricsFromClientSide` reproduces the issue with stock filters (`SingleColumnValueFilter` with `setFilterIfMissing(true)` on a two-family table): before the fix ROWS_SCANNED reported 15 for a scan over 10 rows with 5 matching, after the fix it reports 10. Details: https://issues.apache.org/jira/browse/HBASE-30351 -- 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]
