一、集群启动过程--启动Master

$SPARK_HOME/sbin/start-master.sh

start-master.sh脚本关键内容:

spark-daemon.sh start org.apache.spark.deploy.master.Master 1 --ip $SPARK_MASTER_IP --port $SPARK_MASTER_PORT --webui-port $SPARK_MASTER_WEBUI_PORT

日志信息:$SPARK_HOME/logs/

// :: INFO Remoting: Remoting started; listening on addresses :[akka.tcp://sparkMaster@hadoop000:7077]
// :: INFO master.Master: Starting Spark master at spark://hadoop000:7077
// :: INFO server.Server: jetty-.y.z-SNAPSHOT
// :: INFO server.AbstractConnector: Started SelectChannelConnector@0.0.0.0:
// :: INFO ui.MasterWebUI: Started MasterWebUI at http://hadoop000:8080
// :: INFO master.Master: I have been elected leader! New state: ALIVE

二、集群启动过程--启动Worker

$SPARK_HOME/sbin/start-slaves.sh

start-slaves.sh脚本关键内容:

spark-daemon.sh start org.apache.spark.deploy.worker.Worker master-spark-URL

Worker运行时,需要注册到指定的master url,这里就是spark://hadoop000:7077

Worker启动之后主要做了两件事情:
  1)将自己注册到Master(RegisterWorker);
  2)定期发送心跳信息给Master;

Worker向Master发送注册信息:

Worker.scala
    ==>preStart
      ==>registerWithMaster
        ==>tryRegisterAllMasters
          ==> actor ! RegisterWorker(workerId, host, port, cores, memory, webUi.boundPort, publicAddress)

Master侧收到RegisterWorker通知:

Master.scala
  ==>case RegisterWorker(id, workerHost, workerPort, cores, memory, workerUiPort, publicAddress) => {
      val worker = new WorkerInfo(id, workerHost, workerPort, cores, memory,
    sender, workerUiPort, publicAddress)
    if (registerWorker(worker)) {
      persistenceEngine.addWorker(worker)
        sender ! RegisteredWorker(masterUrl, masterWebUiUrl) //注册成功后向Worker发送注册成功信息
        schedule()
      }
    }

Worker在收到Master发来的注册成功信息后,定期向Master发送心跳信息

Worker.scala
  ==>case SendHeartbeat =>
    masterLock.synchronized {if (connected) { master ! Heartbeat(workerId) }
  }

Master在接收到Worker发送来的心跳信息后更新最后一次心跳时间

Master.scala
  ==>case Heartbeat(workerId) => {
      idToWorker.get(workerId) match {
  case Some(workerInfo) =>
          workerInfo.lastHeartbeat =
System.currentTimeMillis()
      }
  }

Master定期移除超时未发送心跳信息给Master的Worker节点

Master.scala
  ==>preStart
    ==>CheckForWorkerTimeOut
      ==>case CheckForWorkerTimeOut => {timeOutDeadWorkers()} //Check for, and remove, any timed-out workers

日志信息:$SPARK_HOME/logs/

Master部分日志信息:

14/07/22 13:41:36 INFO master.Master: Registering worker hadoop000:48343 with 1 cores, 2.0 GB RAM

Worker部分日志信息:

14/07/22 13:41:35 INFO Worker: Starting Spark worker hadoop000:48343 with 1 cores, 2.0 GB RAM
14/07/22 13:41:35 INFO Worker: Spark home: /home/spark/app/spark-1.0.1-bin-2.3.0-cdh5.0.0
14/07/22 13:41:35 INFO WorkerWebUI: Started WorkerWebUI at http://hadoop000:8081
14/07/22 13:41:35 INFO Worker: Connecting to master spark://hadoop000:7077...
14/07/22 13:41:36 INFO Worker: Successfully registered with master spark://hadoop000:7077

三、Application提交过程

A、提交Application

运行spark-shell: $SPARK_HOME/bin/spark-shell --master spark://hadoop000:7077

日志信息:$SPARK_HOME/work

spark-shell属于application,在启动SparkContext的createTaskScheduler创建SparkDeploySchedulerBackend的过程中创建

client = new AppClient(sc.env.actorSystem, masters, appDesc, this, conf)
client.start()

会向Master发送RegisterApplication请求

AppClient.scala
  ==>preStart
    ==>registerWithMaster
      ==>tryRegisterAllMasters
        ==>actor ! RegisterApplication(appDescription)

B、 Master处理RegisterApplication的请求

在Master侧其处理的分支是RegisterApplication;Master在收到RegisterApplication请求之后,Master进行调度:如果有worker已经注册上来,发送LaunchExecutor指令给相应worker

Master.scala
==>case RegisterApplication(description) => {
logInfo("Registering app " + description.name)
val app = createApplication(description, sender)
registerApplication(app)
logInfo("Registered app " + description.name + " with ID " + app.id)
persistenceEngine.addApplication(app)
sender ! RegisteredApplication(app.id, masterUrl)
schedule()
}
==>schedule
==>launchExecutor(worker, exec)
==> worker.addExecutor(exec)
worker.actor ! LaunchExecutor(masterUrl,exec.application.id, exec.id, exec.application.desc, exec.cores, exec.memory)
exec.application.driver ! ExecutorAdded(exec.id, worker.id, worker.hostPort, exec.cores, exec.memory)

C、启动Executor

Worker在收到LaunchExecutor指令之后,会启动Executor进程

Worker.scala
==>case LaunchExecutor(masterUrl, appId, execId, appDesc, cores_, memory_) =>
logInfo("Asked to launch executor %s/%d for %s".format(appId, execId, appDesc.name))
val manager = new ExecutorRunner(appId, execId, appDesc, cores_, memory_,
self, workerId, host,
appDesc.sparkHome.map(userSparkHome => new File(userSparkHome)).getOrElse(sparkHome),
workDir, akkaUrl, ExecutorState.RUNNING)
executors(appId + "/" + execId) = manager
manager.start()
coresUsed += cores_
memoryUsed += memory_
masterLock.synchronized {master ! ExecutorStateChanged(appId, execId, manager.state, None, None)}
}

D、注册Executor

启动的Executor进程会根据启动时的入参,将自己注册到Driver中的SchedulerBackend

SparkDeploySchedulerBackend.scala
==>preStart (CoarseGrainedSchedulerBackend)
==> case RegisterExecutor(executorId, hostPort, cores) =>
logInfo("Registered executor: " + sender + " with ID " + executorId)
sender ! RegisteredExecutor(sparkProperties)
executorActor(executorId) = sender
executorHost(executorId) = Utils.parseHostPort(hostPort)._1
totalCores(executorId) = cores
freeCores(executorId) = cores
executorAddress(executorId) = sender.path.address
addressToExecutorId(sender.path.address) = executorId
totalCoreCount.addAndGet(cores)
makeOffers() CoarseGrainedExecutorBackend.scala
case RegisteredExecutor(sparkProperties) =>
ogInfo("Successfully registered with driver")
executor = new Executor(executorId, Utils.parseHostPort(hostPort)._1, sparkProperties,false)

executor日志信息位置:控制台/$SPARK_HOME/logs

E、运行Task

示例代码:

sc.textFile("hdfs://hadoop000:8020/hello.txt").flatMap(_.split('\t')).map((_,1)).reduceByKey(_+_).collect

SchedulerBackend收到Executor的注册消息之后,会将提交到的Spark Job分解为多个具体的Task,然后通过LaunchTask指令将这些Task分散到各个Executor上真正的运行

CoarseGrainedSchedulerBackend.scala
def makeOffers() {
launchTasks(scheduler.resourceOffers(
executorHost.toArray.map {case (id, host) => new WorkerOffer(id, host, freeCores(id))}))
} ==>executorActor(task.executorId) ! LaunchTask(new SerializableBuffer(serializedTask))
==>CoarseGrainedSchedulerBackend case LaunchTask(data) =>
if (executor == null) {
logError("Received LaunchTask command but executor was null")
System.exit(1)
} else {
val ser = SparkEnv.get.closureSerializer.newInstance()
val taskDesc = ser.deserialize[TaskDescription](data.value)
logInfo("Got assigned task " + taskDesc.taskId)
executor.launchTask(this, taskDesc.taskId, taskDesc.serializedTask)
}

Master部分日志信息:

14/07/22 15:25:27 INFO master.Master: Registering app Spark shell
14/07/22 15:25:27 INFO master.Master: Registered app Spark shell with ID app-20140722152527-0001
14/07/22 15:25:27 INFO master.Master: Launching executor app-20140722152527-0001/0 on worker worker-20140722134135-hadoop000-48343

Worker部分日志信息:

Spark assembly has been built with Hive, including Datanucleus jars on classpath
14/07/22 15:25:27 INFO Worker: Asked to launch executor app-20140722152527-0001/0 for Spark shell
Spark assembly has been built with Hive, including Datanucleus jars on classpath
14/07/22 15:25:28 INFO ExecutorRunner: Launch command: "java" "-cp" "::/home/spark/app/spark-1.0.1-bin-2.3.0-cdh5.0.0/conf:/home/spark/app/spark-1.0.1-bin-2.3.0-cdh5.0.0/lib/spark-assembly-1.0.1-hadoop2.3.0-cdh5.0.0.jar:/home/spark/app/spark-1.0.1-bin-2.3.0-cdh5.0.0/lib/datanucleus-rdbms-3.2.1.jar:/home/spark/app/spark-1.0.1-bin-2.3.0-cdh5.0.0/lib/datanucleus-core-3.2.2.jar:/home/spark/app/spark-1.0.1-bin-2.3.0-cdh5.0.0/lib/datanucleus-api-jdo-3.2.1.jar" "-XX:MaxPermSize=128m" "-Xms1024M" "-Xmx1024M" "org.apache.spark.executor.CoarseGrainedExecutorBackend" "akka.tcp://spark@hadoop000:50515/user/CoarseGrainedScheduler" "0" "hadoop000" "1" "akka.tcp://sparkWorker@hadoop000:48343/user/Worker" "app-20140722152527-0001"

控制台部分日志信息:

14/07/22 15:25:31 INFO cluster.SparkDeploySchedulerBackend: Registered executor: Actor[akka.tcp://sparkExecutor@hadoop000:45150/user/Executor#-791712793] with ID 0
14/07/22 15:25:31 INFO CoarseGrainedExecutorBackend: Successfully registered with driver

每当有新的application注册到master,master都要调度schedule函数将application发送到相应的worker,在对应的worker启动相应的ExecutorBackend,最终的Task就运行在ExecutorBackend中

Spark分析之Standalone运行过程分析的更多相关文章

  1. Spark分析之SparkContext启动过程分析

    SparkContext作为整个Spark的入口,不管是spark.sparkstreaming.spark sql都需要首先创建一个SparkContext对象,然后基于这个SparkContext ...

  2. Spark standalone运行模式

    Spark Standalone 部署配置 Standalone架构 手工启动一个Spark集群 https://spark.apache.org/docs/latest/spark-standalo ...

  3. Task的运行过程分析

    Task的运行过程分析 Task的运行通过Worker启动时生成的Executor实例进行, caseRegisteredExecutor(sparkProperties)=> logInfo( ...

  4. 【Spark Core】任务运行机制和Task源代码浅析1

    引言 上一小节<TaskScheduler源代码与任务提交原理浅析2>介绍了Driver側将Stage进行划分.依据Executor闲置情况分发任务,终于通过DriverActor向exe ...

  5. [大数据从入门到放弃系列教程]第一个spark分析程序

    [大数据从入门到放弃系列教程]第一个spark分析程序 原文链接:http://www.cnblogs.com/blog5277/p/8580007.html 原文作者:博客园--曲高终和寡 **** ...

  6. Spark新手入门——3.Spark集群(standalone模式)安装

    主要包括以下三部分,本文为第三部分: 一. Scala环境准备 查看二. Hadoop集群(伪分布模式)安装 查看三. Spark集群(standalone模式)安装 Spark集群(standalo ...

  7. 五、standalone运行模式

    在上文中我们知道spark的集群主要有三种运行模式standalone.yarn.mesos,其中常被使用的是standalone和yarn,本文了解一下什么是standalone运行模式,它的运行流 ...

  8. Netty3 源代码分析 - NIO server绑定过程分析

    Netty3 源代码分析 - NIO server绑定过程分析      一个框架封装的越好,越利于我们高速的coding.可是却掩盖了非常多的细节和原理.可是源代码可以揭示一切. 服务器端代码在指定 ...

  9. Spark 中 RDD的运行机制

    1. RDD 的设计与运行原理 Spark 的核心是建立在统一的抽象 RDD 之上,基于 RDD 的转换和行动操作使得 Spark 的各个组件可以无缝进行集成,从而在同一个应用程序中完成大数据计算任务 ...

随机推荐

  1. Redis学习第五课:Redis Set类型及操作

    Set是集合,它是string类型的无序集合.set是通过hash table实现的,添加.删除和查找的复杂度都是O(1). 对集合我们可以取并集.交集.差集.通过这些操作我们可以实现SNS中的好友推 ...

  2. antd中fomr中resetFields清空输入框

    1.如果没有initValue的情况下,直接使用resetFields可以清空文本框的值 2.如果是有initValue的情况下,直接使用resetFields方法会直接重置为initValue的值 ...

  3. Mac 活动监视器 闪退 pro发热耗电过快问题解决,亲测可用解决

    该解决办法转载 Mac 活动监视器 闪退 pro发热耗电过快问题解决 这个月新买了mac,升级了系统,出现CPU发热,高负荷运转问题,始终找不到问题解决办法, ,这个过程太痛苦了,也不知道是什么原因. ...

  4. magento -- 给Magento提速之缓存上的探索

    依然在为Magento提速做努力,除了自带的缓存和编译,之前的所作的很多努力都是从减少JS,Css,图片等载入时间入手,而对页面载入耗时最早有时也是最大的一部分--获取页面数据没有做太多处理,以gap ...

  5. OK335xS U-boot GPIO control hacking

    /**************************************************************************************** * OK335xS ...

  6. OK335xS GPMC nand device register hacking

    /********************************************************************************* * OK335xS GPMC na ...

  7. GPIO口的输入输出模式

    1.浮空输入  GPIO_Mode_IN_FLOATING       2.带上拉输入  GPIO_Mode_IPU       3.带下拉输入  GPIO_Mode_IPD       4.模拟输入 ...

  8. java8 array、list操作 汇【2】)- (Function,Consumer,Predicate,Supplier)应用

    static class UserT { String name; public UserT(String zm) { this.name=zm; } public String getName() ...

  9. python 常见的内置函数

    内置函数 接下来,我们就一起来看看python里的内置函数.截止到python版本3.6.2,现在python一共为我们提供了68个内置函数.它们就是python提供给你直接可以拿来使用的所有函数.这 ...

  10. oracle mysql sql serve where in 语句的不同

    类似这样的语句在mysql  oracle 是可以执行成功的, select * from classfirst where (classid ,classname) not in (select c ...