You are viewing a plain text version of this content. The canonical link for it is here.
Posted to reviews@spark.apache.org by GitBox <gi...@apache.org> on 2022/01/06 17:54:21 UTC
[GitHub] [spark] venkata91 commented on a change in pull request #34122: [SPARK-34826][SHUFFLE] Adaptively fetch shuffle mergers for push based shuffle
venkata91 commented on a change in pull request #34122:
URL: https://github.com/apache/spark/pull/34122#discussion_r779745057
##########
File path: core/src/main/scala/org/apache/spark/Dependency.scala
##########
@@ -144,12 +144,16 @@ class ShuffleDependency[K: ClassTag, V: ClassTag, C: ClassTag](
_shuffleMergedFinalized = true
}
+ def shuffleMergeFinalized: Boolean = {
Review comment:
Yes I agree. With Adaptively fetching shuffle mergers, we need to check whether the `shuffleId` is `shuffleMergeFinalized` or not but that internally checks `shuffleMergeEnabled` which would be false since we didn't have enough mergers during the start of the stage. Which is why we need to have an additional method distinguishing whether shuffle merge is finalized vs whether shuffleMergeEnabled and then shuffleMergeFinalized whenever we try to figure which stage to be computed next. We need to prevent the next stage from starting if the previous stage shuffle merge finalization is yet to be completed ([here](https://github.com/apache/spark/blob/f6128a6f4215dc45a19209d799dd9bf98fab6d8a/core/src/main/scala/org/apache/spark/scheduler/DAGScheduler.scala#L731)).
--
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