venkata91 commented on a change in pull request #34122:
URL: https://github.com/apache/spark/pull/34122#discussion_r787356062
##########
File path:
core/src/test/scala/org/apache/spark/scheduler/DAGSchedulerSuite.scala
##########
@@ -4147,7 +4146,128 @@ class DAGSchedulerSuite extends SparkFunSuite with
TempLocalSparkContext with Ti
assert(finalizeTask2.delay == 0 && finalizeTask2.registerMergeResults)
}
- /**
+ test("SPARK-34826: Adaptively fetch shuffle mergers") {
+ initPushBasedShuffleConfs(conf)
+ conf.set(config.SHUFFLE_MERGER_LOCATIONS_MIN_STATIC_THRESHOLD, 6)
+ DAGSchedulerSuite.clearMergerLocs
+ DAGSchedulerSuite.addMergerLocs(Seq("host1", "host2", "host3", "host4",
"host5"))
+ val parts = 7
+
+ val shuffleMapRdd = new MyRDD(sc, parts, Nil)
+ val shuffleDep = new ShuffleDependency(shuffleMapRdd, new
HashPartitioner(parts))
+ val reduceRdd = new MyRDD(sc, parts, List(shuffleDep), tracker =
mapOutputTracker)
+
+ // Submit a reduce job that depends which will create a map stage
+ submit(reduceRdd, (0 until parts).toArray)
+
+ runEvent(makeCompletionEvent(
+ taskSets(0).tasks(0), Success, makeMapStatus("hostA", parts),
+ Seq.empty, Array.empty, createFakeTaskInfoWithId(0)))
+
+ val shuffleStage1 =
scheduler.stageIdToStage(0).asInstanceOf[ShuffleMapStage]
+ assert(shuffleStage1.shuffleDep.getMergerLocs.isEmpty)
+ assert(mapOutputTracker.getShufflePushMergerLocations(0).isEmpty)
+
+ DAGSchedulerSuite.addMergerLocs(Seq("host6", "host7", "host8"))
+
+ runEvent(makeCompletionEvent(
+ taskSets(0).tasks(1), Success, makeMapStatus("hostA", parts),
+ Seq.empty, Array.empty, createFakeTaskInfoWithId(1)))
Review comment:
Yes, this is not required.
` runEvent(makeCompletionEvent(
taskSets(0).tasks(1), Success, makeMapStatus("hostA", parts),
Seq.empty, Array.empty, createFakeTaskInfoWithId(1)))`
I probably must have written the test thinking that the `mergerLocs` are
fetched from tasks rather it is pushed from `DAGScheduler`. Must be some
confusion while writing the test. Clearly it is not required.
--
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]
