[GitHub] [incubator-celeborn] waitinfuture commented on a diff in pull request #1444: [CELEBORN-541][PERF] handleGetReducerFileGroup occupy too much RPC thread cause other RPC can't been handled

["waitinfuture (via GitHub)" <gi...@apache.org>]

waitinfuture commented on code in PR #1444:
URL: https://github.com/apache/incubator-celeborn/pull/1444#discussion_r1173430611


##########
client/src/main/scala/org/apache/celeborn/client/commit/ReducePartitionCommitHandler.scala:
##########
@@ -93,7 +95,11 @@ class ReducePartitionCommitHandler(
   }
 
   override def setStageEnd(shuffleId: Int): Unit = {
-    stageEndShuffleSet.add(shuffleId)
+    getReducerFileGroupRequest synchronized {
+      stageEndShuffleSet.add(shuffleId)
+      getReducerFileGroupRequest.remove(shuffleId)
+        .asScala.foreach(replyGetReducerFileGroup(_, shuffleId))

Review Comment:
   IMO we should move the invocation of replyGetReducerFileGroup out the synchronize block to avoid the critical section cost too much time.



-- 
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: issues-unsubscribe@celeborn.apache.org

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