我的Spark源码核心SparkContext走读全纪录

我的Spark源码核心SparkContext走读全纪录

专注于为中小企业提供成都网站建设、成都网站制作服务,电脑端+手机端+微信端的三站合一,更高效的管理,为中小企业三门峡免费做网站提供优质的服务。我们立足成都，凝聚了一批互联网行业人才，有力地推动了近1000家企业的稳健成长，帮助中小企业通过网站建设实现规模扩充和转变。

Dirver Program(SparkConf)  package org.apache.spark.SparkConf

Master        package org.apache.spark.deploy.master

SparkContext  package org.apache.spark.SparkContext

Stage         package org.apache.spark.scheduler.Stage

Task          package org.apache.spark.scheduler.Task  

DAGScheduler  package org.apache.spark.scheduler   

TaskScheduler package org.apache.spark.scheduler.TaskScheduler

TaskSchedulerImpl  package org.apache.spark.scheduler

Worker        package org.apache.spark.deploy.worker

Executor      package org.apache.spark.executor

BlockManager  package org.apache.spark.storage

TaskSet       package org.apache.spark.scheduler

//初始化后开始创建

// Create and start the scheduler

    val (sched, ts) = SparkContext.createTaskScheduler(this, master)

    _schedulerBackend = sched

    _taskScheduler = ts

    _dagScheduler = new DAGScheduler(this)

    _heartbeatReceiver.send(TaskSchedulerIsSet)

/**

   * Create a task scheduler based on a given master URL.

   * Return a 2-tuple of the scheduler backend and the task scheduler.

   */

  private def createTaskScheduler(

      sc: SparkContext,

      master: String): (SchedulerBackend, TaskScheduler) = {

master match {

      case "local" =>

实例化一个

val scheduler = new TaskSchedulerImpl(sc)

构建masterUrls：

val masterUrls = localCluster.start()

据说是非常关键的backend：

val backend = new SparkDeploySchedulerBackend(scheduler, sc, masterUrls)

        scheduler.initialize(backend)

        backend.shutdownCallback = (backend: SparkDeploySchedulerBackend) => {

          localCluster.stop()

        }

        (backend, scheduler)