上篇文章《

Spark 源码解析 : DAGScheduler中的DAG划分与提交

》介绍了DAGScheduler的Stage划分算法。

本文继续分析Stage被封装成TaskSet，并将TaskSet提交到集群的Executor执行的过程

在DAGScheduler的submitStage方法中，将Stage划分完成，生成拓扑结构，当一个stage没有父stage时候，会调用DAGScheduler的submitMissingTasks方法来提交该stage包含tasks。

首先来分析一下DAGScheduler的submitMissingTasks方法

1.获取Task的最佳计算位置：

    val taskIdToLocations: Map[Int, Seq[TaskLocation]] = try {
      stage match {
        case s: ShuffleMapStage =>
          partitionsToCompute.map { id => (id, getPreferredLocs(stage.rdd, id))}.toMap
        case s: ResultStage =>
          val job = s.activeJob.get
          partitionsToCompute.map { id =>
            val p = s.partitions(id)
            (id, getPreferredLocs(stage.rdd, p))
          }.toMap
      }
    }

核心是其中的getPreferredLocs方法，根据RDD的数据信息得到task的最佳计算位置，从而获取较好的数据本地性。其中的细节这里先跳过，在以后的文章在做分析

2.序列化Task的Binary，并进行广播。Executor端在执行task时会向反序列化Task。

3.根据stage的不同类型创建，为stage的每个分区创建创建task,并封装成TaskSet。Stage分两种类型ShuffleMapStage生成ShuffleMapTask，ResultStage生成ResultTask。

 val tasks: Seq[Task[_]] = try {
      stage match {
        case stage: ShuffleMapStage =>
          partitionsToCompute.map { id =>
            val locs = taskIdToLocations(id)
            val part = stage.rdd.partitions(id)
            new ShuffleMapTask(stage.id, stage.latestInfo.attemptId,
              taskBinary, part, locs, stage.internalAccumulators)
          }

        case stage: ResultStage =>
          val job = stage.activeJob.get
          partitionsToCompute.map { id =>
            val p: Int = stage.partitions(id)
            val part = stage.rdd.partitions(p)
            val locs = taskIdToLocations(id)
            new ResultTask(stage.id, stage.latestInfo.attemptId,
              taskBinary, part, locs, id, stage.internalAccumulators)
          }
      }

4.调用TaskScheduler的submitTasks，提交TaskSet

      logInfo("Submitting " + tasks.size + " missing tasks from " + stage + " (" + stage.rdd + ")")
      stage.pendingPartitions ++= tasks.map(_.partitionId)
      logDebug("New pending partitions: " + stage.pendingPartitions)
      taskScheduler.submitTasks(new TaskSet(
        tasks.toArray, stage.id, stage.latestInfo.attemptId, jobId, properties))
      stage.latestInfo.submissionTime = Some(clock.getTimeMillis())

submitTasks方法的实现在TaskScheduler的实现类TaskSchedulerImpl中。

4.1 TaskSchedulerImpl的submitTasks方法首先创建TaskSetManager。

val manager = createTaskSetManager(taskSet, maxTaskFailures)
      val stage = taskSet.stageId
      val stageTaskSets =
        taskSetsByStageIdAndAttempt.getOrElseUpdate(stage, new HashMap[Int, TaskSetManager])
      stageTaskSets(taskSet.stageAttemptId) = manager

TaskSetManager负责管理TaskSchedulerImpl中一个单独TaskSet，跟踪每一个task，如果task失败，负责重试task直到达到task重试次数的最多次数。并且通过延迟调度来执行task的位置感知调度。

private[spark] class TaskSetManager(
    sched: TaskSchedulerImpl,//绑定的TaskSchedulerImpl
    val taskSet: TaskSet,
    val maxTaskFailures: Int, //失败最大重试次数
    clock: Clock = new SystemClock())
  extends Schedulable with Logging

4.2 将TaskSetManger加入schedulableBuilder

schedulableBuilder.addTaskSetManager(manager, manager.taskSet.properties) //将TaskSetManager加入rootPool调度池中，由schedulableBuilder决定调度顺序

schedulableBuilder的类型是 SchedulerBuilder，SchedulerBuilder是一个trait，有两个实现FIFOSchedulerBuilder和 FairSchedulerBuilder，并且默认采用的是FIFO方式

  // default scheduler is FIFO
  private val schedulingModeConf = conf.get("spark.scheduler.mode", "FIFO")

而schedulableBuilder的创建是在SparkContext创建SchedulerBackend和TaskScheduler后调用TaskSchedulerImpl的初始化方法进行创建的。

  def initialize(backend: SchedulerBackend) {
    this.backend = backend
    // temporarily set rootPool name to empty
    rootPool = new Pool("", schedulingMode, 0, 0)
    schedulableBuilder = {
      schedulingMode match {
        case SchedulingMode.FIFO =>
          new FIFOSchedulableBuilder(rootPool)
        case SchedulingMode.FAIR =>
          new FairSchedulableBuilder(rootPool, conf)
      }
    }
    schedulableBuilder.buildPools()
  }

schedulableBuilder是TaskScheduler中一个重要成员，他根据调度策略决定了TaskSetManager的调度顺序。

4.3 接下来调用SchedulerBackend的riviveOffers方法对Task进行调度，决定task具体运行在哪个Executor中。

调用CoarseGrainedSchedulerBackend的riviveOffers方法，该方法给driverEndpoint发送ReviveOffer消息

  override def reviveOffers() {
    driverEndpoint.send(ReviveOffers)
  }

driverEndpoint收到ReviveOffer消息后调用makeOffers方法

    // Make fake resource offers on all executors
    private def makeOffers() {
      //过滤出活跃状态的Executor
      val activeExecutors = executorDataMap.filterKeys(executorIsAlive)
      //将Executor封装成WorkerOffer对象
      val workOffers = activeExecutors.map { case (id, executorData) =>
        new WorkerOffer(id, executorData.executorHost, executorData.freeCores)
      }.toSeq

      launchTasks(scheduler.resourceOffers(workOffers))
    }

注意：上面代码中的executorDataMap，在客户的向Master注册Application的时候，Master已经为Application分配并启动好Executor，然后注册给CoarseGrainedSchedulerBackend，注册信息就是存储在executorDataMap数据结构中。

准备好计算资源后，接下来TaskSchedulerImpl基于这些计算资源为task分配Executor。

我们看一下TaskSchedulerImpl的resourceOffers方法：

   // 随机打乱offers
    val shuffledOffers = Random.shuffle(offers)

    // 构建一个二维数组，保存每个Executor上将要分配的那些task
    val tasks = shuffledOffers.map(o => new ArrayBuffer[TaskDescription](o.cores))
    val availableCpus = shuffledOffers.map(o => o.cores).toArray


    
    //根据SchedulerBuilder的调度算法，给TaskManager排好序

val sortedTaskSets = rootPool.getSortedTaskSetQueue

    for (taskSet <- sortedTaskSets) {
      logDebug("parentName: %s, name: %s, runningTasks: %s".format(
        taskSet.parent.name, taskSet.name, taskSet.runningTasks))
      if (newExecAvail) {
        taskSet.executorAdded()
      }
    }

    // 使用双重循环，对每一个taskset 依照调度的顺序，依次按照本地性级别顺序尝试启动task
    // 数据本地性级别顺序: PROCESS_LOCAL, NODE_LOCAL, NO_PREF, RACK_LOCAL, ANY
    var launchedTask = false
    for (taskSet <- sortedTaskSets; maxLocality <- taskSet.myLocalityLevels) {
      do {
        launchedTask = resourceOfferSingleTaskSet(
            taskSet, maxLocality, shuffledOffers, availableCpus, tasks)
      } while (launchedTask)
    }

    if (tasks.size > 0) {
      hasLaunchedTask = true
    }
    return tasks

下面看看 resourceOfferSingleTaskSet 方法：

用当前的数据本地性，调用TaskSetManager的resourceOffer方法，在当前executor上分配task

private def resourceOfferSingleTaskSet(
      taskSet: TaskSetManager,
      maxLocality: TaskLocality,
      shuffledOffers: Seq[WorkerOffer],
      availableCpus: Array[Int],
      tasks: Seq[ArrayBuffer[TaskDescription]]) : Boolean = {
    var launchedTask = false
    for (i <- 0 until shuffledOffers.size) {
      val execId = shuffledOffers(i).executorId
      val host = shuffledOffers(i).host
        //如果executor 的cup数大于 每个task的cup数目（值为1）
      if (availableCpus(i) >= CPUS_PER_TASK) {
        try {
        //
          for (task <- taskSet.resourceOffer(execId, host, maxLocality)) {
            tasks(i) += task
            val tid = task.taskId
            taskIdToTaskSetManager(tid) = taskSet
            taskIdToExecutorId(tid) = execId
            executorIdToTaskCount(execId) += 1
            executorsByHost(host) += execId
            availableCpus(i) -= CPUS_PER_TASK
            assert(availableCpus(i) >= 0)
            launchedTask = true
          }
        }

为Task分配好资源之后，DriverEndpint调用launchTask方法将task在Executor上启动运行。task在Executor上的启动运行过程，在后面的文章中会继续分析，敬请关注。

总结一下调用过程：

TaskSchedulerImpl#submitTasks

CoarseGrainedSchedulerBackend#riviveOffers

CoarseGrainedSchedulerBackend$DriverEndpoint#makeOffers

|-TaskSchedulerImpl#resourceOffers(offers) 为offers分配task

|- TaskSchedulerImpl#resourceOfferSingleTaskSet

CoarseGrainedSchedulerBackend$DriverEndpoint#launchTask

来自为知笔记(Wiz)