otterc commented on code in PR #37922:
URL: https://github.com/apache/spark/pull/37922#discussion_r1067270476
##########
core/src/main/scala/org/apache/spark/storage/BlockManagerMasterEndpoint.scala:
##########
@@ -321,6 +321,12 @@ class BlockManagerMasterEndpoint(
}
private def removeShuffle(shuffleId: Int): Future[Seq[Boolean]] = {
+ val mergerLocations =
+ if (Utils.isPushBasedShuffleEnabled(conf, isDriver)) {
+ mapOutputTracker.getShufflePushMergerLocations(shuffleId)
+ } else {
+ Seq.empty[BlockManagerId]
+ }
Review Comment:
IIUC below sends the message to all the executors for removing shuffle
```
val removeMsg = RemoveShuffle(shuffleId)
val removeShuffleFromExecutorsFutures = blockManagerInfo.values.map { bm
=>
bm.storageEndpoint.ask[Boolean](removeMsg).recover {
// use false as default value means no shuffle data were removed
handleBlockRemovalFailure("shuffle", shuffleId.toString,
bm.blockManagerId, false)
}
}.toSeq
```
In the `BlockManagerStorageEndpoint`, the below will remove the statuses
cached in the executors and not the driver.
case RemoveShuffle(shuffleId) =>
doAsync[Boolean]("removing shuffle " + shuffleId, context) {
if (mapOutputTracker != null) {
mapOutputTracker.unregisterShuffle(shuffleId)
}
SparkEnv.get.shuffleManager.unregisterShuffle(shuffleId)
}
So, I don't think this should impact the implementation of `removeShuffle`
in the `BlockManagerMasterEndpoint`.
cc. @mridulm
--
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]
