sarutak commented on PR #56595: URL: https://github.com/apache/spark/pull/56595#issuecomment-4754109661
Thank you for the feedback, @peter-toth @uros-b. Added a test that forces spill by setting `spark.sql.sortMergeJoinExec.buffer.in.memory.threshold` and `spark.sql.sortMergeJoinExec.buffer.spill.threshold` to `1`. The test covers both the no-equi-key path (bufferAllRight) and the equi-key path (bufferRightGroup). Regarding the I/O concern: the per-left-row generateIterator() re-scan is the same pattern as `SortMergeJoinExec` (which also re-iterates bufferedMatches for each streamed row). In practice, spill only occurs for extremely skewed groups where the alternative was OOM. The typical case stays in-memory. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
