[GitHub] [spark] Hisoka-X commented on a diff in pull request #42194: [SPARK-41471][SQL] Reduce Spark shuffle when only one side of a join is KeyGroupedPartitioning

["Hisoka-X (via GitHub)" <gi...@apache.org>]

Hisoka-X commented on code in PR #42194:
URL: https://github.com/apache/spark/pull/42194#discussion_r1287859433


##########
core/src/main/scala/org/apache/spark/Partitioner.scala:
##########
@@ -137,6 +137,18 @@ private[spark] class PartitionIdPassthrough(override val numPartitions: Int) ext
   override def getPartition(key: Any): Int = key.asInstanceOf[Int]
 }
 
+/**
+ * A [[org.apache.spark.Partitioner]] that partitions all records use partition value map
+ */
+private[spark] class PartitionValueMapPartitioner(
+    valueMap: mutable.Map[Seq[Any], Int],
+    override val numPartitions: Int) extends Partitioner {
+  override def getPartition(key: Any): Int = {
+    val keys = key.asInstanceOf[Seq[Any]]
+    valueMap.getOrElseUpdate(keys, Utils.nonNegativeMod(keys.hashCode, numPartitions))

Review Comment:
   > When do we need to update this map?
   
   When the shuffle side (without KeyGroupedPartitioning side) have some partition value not in KeyGroupedPartitioning. This is a fallback operation to ensure that all data will be shuffled. In fact we can add Filter to avoid this kind data to be shuffled. We can do it later.
   
   > will this collide with one of the existing mappings?
   
   no, just update not existed key.



-- 
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: reviews-unsubscribe@spark.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org