ulysses-you commented on a change in pull request #32816:
URL: https://github.com/apache/spark/pull/32816#discussion_r703278537
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/adaptive/OptimizeSkewedJoin.scala
##########
@@ -250,15 +253,28 @@ object OptimizeSkewedJoin extends AQEShuffleReadRule {
// SHJ
// Shuffle
// Shuffle
- optimizeSkewJoin(plan)
+ val optimized =
+ EnsureRequirements(requiredDistribution.isDefined,
requiredDistribution)
+ .apply(optimizeSkewJoin(plan))
+ val originCost = costEvaluator.evaluateCost(plan)
+ val optimizedCost = costEvaluator.evaluateCost(optimized)
+ // two cases we will pick new plan:
+ // 1. optimize the skew join without extra shuffle
+ // 2. optimize the skew join with extra shuffle but the costEvaluator
think it's better
+ if (optimizedCost < originCost || (originCost == optimizedCost &&
optimized != plan)) {
Review comment:
yea, since we are only optimize physical plan here
--
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]
