tcondie commented on issue #25491: [SPARK-28699][SQL] Disable using radix sort for ShuffleExchangeExec in repartition case URL: https://github.com/apache/spark/pull/25491#issuecomment-551234051 @xuanyuanking could you please clarify your comment: ``` // We are comparing binary here, which does not support radix sort. // See more details in SPARK-28699. ``` When the code seems to be sorting on the hashCode of the row: ``` // The comparator for comparing row hashcode, which should always be Integer. val prefixComparator = PrefixComparators.LONG ``` After rolling this fix into HDInsight Spark, we are still seeing non-deterministic behavior in round-robin shuffles. My hunch is that it may have something to do with row hashCode collisions but I have thus far not been able to produce a repro that can be shared externally i.e., this is a customer facing issue.
---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
