[jira] [Commented] (SPARK-22593) submitMissingTask in DagScheduler will call partitions function many times whch may be time consuming

["Hyukjin Kwon (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/SPARK-22593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16268370#comment-16268370 ] 

Hyukjin Kwon commented on SPARK-22593:
--------------------------------------

Do you maybe have some measure for the elapsed time? In a quick look, it looks it's trivial and a non-critical path. To be honest, I won't bother it if it's true.

Is it a question or an issue?

> submitMissingTask in DagScheduler will call partitions function many times whch may be time consuming
> -----------------------------------------------------------------------------------------------------
>
>                 Key: SPARK-22593
>                 URL: https://issues.apache.org/jira/browse/SPARK-22593
>             Project: Spark
>          Issue Type: Question
>          Components: Spark Core
>    Affects Versions: 2.2.0
>            Reporter: tomzhu
>            Priority: Minor
>
> when dagScheduler call submitMissing task, will create tasks and calling stage.rdd.partitions, it will can many times which may be time-consuming, the code is:
> {quote}
>     val tasks: Seq[Task[_]] = try {
>       val serializedTaskMetrics = closureSerializer.serialize(stage.latestInfo.taskMetrics).array()
>       stage match {
>         case stage: ShuffleMapStage =>
>           stage.pendingPartitions.clear()
>           partitionsToCompute.map { id =>
>             val locs = taskIdToLocations(id)
>             val part = stage.rdd.partitions(id)
>             stage.pendingPartitions += id
>             new ShuffleMapTask(stage.id, stage.latestInfo.attemptId,
>               taskBinary, part, locs, properties, serializedTaskMetrics, Option(jobId),
>               Option(sc.applicationId), sc.applicationAttemptId)
>           }
>         case stage: ResultStage =>
>           partitionsToCompute.map { id =>
>             val p: Int = stage.partitions(id)
>             val part = stage.rdd.partitions(p)  //here is a little time  consuming.
>             val locs = taskIdToLocations(id)
>             new ResultTask(stage.id, stage.latestInfo.attemptId,
>               taskBinary, part, locs, id, properties, serializedTaskMetrics,
>               Option(jobId), Option(sc.applicationId), sc.applicationAttemptId)
>           }
>       }
>     } 
> {quote}
> for example, for a parallelCollectionRdd with 3 slices or partitions, to create task, the code will call stage.rdd.partitions three times, since stage.rdd.partitions will call getPartitions, so getPartions will call three times, it is a little time-cousuming. the stage.rdd.partitions code :
> {quote}  
> final def partitions: Array[Partition] = {
>     checkpointRDD.map(_.partitions).getOrElse {
>       if (partitions_ == null) {
>         partitions_ = getPartitions
>         partitions_.zipWithIndex.foreach { case (partition, index) =>
>           require(partition.index == index,
>             s"partitions($index).partition == ${partition.index}, but it should equal $index")
>         }
>       }
>       partitions_
>     }
>   }
> {quote}
> it would be better to avoid this.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org