venkata91 commented on a change in pull request #34122:
URL: https://github.com/apache/spark/pull/34122#discussion_r780451587
##########
File path: core/src/main/scala/org/apache/spark/scheduler/StageInfo.scala
##########
@@ -39,7 +39,9 @@ class StageInfo(
val taskMetrics: TaskMetrics = null,
private[spark] val taskLocalityPreferences: Seq[Seq[TaskLocation]] =
Seq.empty,
private[spark] val shuffleDepId: Option[Int] = None,
- val resourceProfileId: Int) {
+ val resourceProfileId: Int,
+ private[spark] var isPushBasedShuffleEnabled: Boolean = false,
+ private[spark] var shuffleMergerCount: Int = 0) {
Review comment:
Yes, currently there are no consumers with in Spark. But
[[SPARK-36620][SHUFFLE] Add client side push based shuffle
metrics](https://github.com/apache/spark/pull/34000) has to use these metrics
and populate it which is currently ongoing. Do you think we can move this
change as well to that PR or wait for this one to get merged and then
correspondingly make the changes in the other PR to consume these metrics? cc
@thejdeep
##########
File path: core/src/main/scala/org/apache/spark/scheduler/DAGScheduler.scala
##########
@@ -2487,6 +2501,21 @@ private[spark] class DAGScheduler(
executorFailureEpoch -= execId
}
shuffleFileLostEpoch -= execId
+
+ if (pushBasedShuffleEnabled) {
Review comment:
Agree. Let me see how we can send an event when a new blockManager's
host which is not already part of existing shuffleMergers and correspondingly
fire an event to DAGScheduler
--
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]
