spark通信流程

概述

spark作为一套高效的分布式运算框架，但是想要更深入的学习它，就要通过分析spark的源码，不但可以更好的帮助理解spark的工作过程，还可以提高对集群的排错能力，本文主要关注的是Spark的Master的启动流程与Worker启动流程。

现在Spark最新版本为1.6，但是代码的逻辑不够清晰，不便于理解，这里以1.3为准

Master启动

我们启动一个Master是通过Shell命令启动了一个脚本start-master.sh开始的，这个脚本的启动流程如下

start-master.sh  -> spark-daemon.sh start org.apache.spark.deploy.master.Master

我们可以看到脚本首先启动了一个org.apache.spark.deploy.master.Master类，启动时会传入一些参数，比如cpu的执行核数，内存大小，app的main方法等

查看Master类的main方法

private[spark] object Master extends Logging {
  val systemName = "sparkMaster"
  private val actorName = "Master"

  //master启动的入口
  def main(argStrings: Array[String]) {
    SignalLogger.register(log)
    //创建SparkConf
    val conf = new SparkConf
    //保存参数到SparkConf
    val args = new MasterArguments(argStrings, conf)
    //创建ActorSystem和Actor
    val (actorSystem, _, _, _) = startSystemAndActor(args.host, args.port, args.webUiPort, conf)
    //等待结束
    actorSystem.awaitTermination()
  }

这里主要看startSystemAndActor方法

  /**
   * Start the Master and return a four tuple of:
   *   (1) The Master actor system
   *   (2) The bound port
   *   (3) The web UI bound port
   *   (4) The REST server bound port, if any
   */
  def startSystemAndActor(
      host: String,
      port: Int,
      webUiPort: Int,
      conf: SparkConf): (ActorSystem, Int, Int, Option[Int]) = {
    val securityMgr = new SecurityManager(conf)

    //利用AkkaUtils创建ActorSystem
    val (actorSystem, boundPort) = AkkaUtils.createActorSystem(systemName, host, port, conf = conf,
      securityManager = securityMgr)
   
    val actor = actorSystem.actorOf(
      Props(classOf[Master], host, boundPort, webUiPort, securityMgr, conf), actorName)
   ....
  }
}

spark底层通信使用的是Akka
通过ActorSystem创建Actor -> actorSystem.actorOf, 就会执行Master的构造方法->然后执行Actor生命周期方法
执行Master的构造方法初始化一些变量

 private[spark] class Master(
    host: String,
    port: Int,
    webUiPort: Int,
    val securityMgr: SecurityManager,
    val conf: SparkConf)
  extends Actor with ActorLogReceive with Logging with LeaderElectable {
  //主构造器

  //启用定期器功能
  import context.dispatcher   // to use Akka's scheduler.schedule()

  val hadoopConf = SparkHadoopUtil.get.newConfiguration(conf)

  def createDateFormat = new SimpleDateFormat("yyyyMMddHHmmss")  // For application IDs
  //woker超时时间
  val WORKER_TIMEOUT = conf.getLong("spark.worker.timeout", 60) * 1000
  val RETAINED_APPLICATIONS = conf.getInt("spark.deploy.retainedApplications", 200)
  val RETAINED_DRIVERS = conf.getInt("spark.deploy.retainedDrivers", 200)
  val REAPER_ITERATIONS = conf.getInt("spark.dead.worker.persistence", 15)
  val RECOVERY_MODE = conf.get("spark.deploy.recoveryMode", "NONE")

  //一个HashSet用于保存WorkerInfo
  val workers = new HashSet[WorkerInfo]
  //一个HashMap用保存workid -> WorkerInfo
  val idToWorker = new HashMap[String, WorkerInfo]
  val addressToWorker = new HashMap[Address, WorkerInfo]

  //一个HashSet用于保存客户端（SparkSubmit）提交的任务
  val apps = new HashSet[ApplicationInfo]
  //一个HashMap Appid-》 ApplicationInfo
  val idToApp = new HashMap[String, ApplicationInfo]
  val actorToApp = new HashMap[ActorRef, ApplicationInfo]
  val addressToApp = new HashMap[Address, ApplicationInfo]
  //等待调度的App
  val waitingApps = new ArrayBuffer[ApplicationInfo]
  val completedApps = new ArrayBuffer[ApplicationInfo]
  var nextAppNumber = 0
  val appIdToUI = new HashMap[String, SparkUI]

  //保存DriverInfo
  val drivers = new HashSet[DriverInfo]
  val completedDrivers = new ArrayBuffer[DriverInfo]
  val waitingDrivers = new ArrayBuffer[DriverInfo] // Drivers currently spooled for scheduling

主构造器执行完就会执行preStart --》执行完receive方法

  //启动定时器，进行定时检查超时的worker
  //重点看一下CheckForWorkerTimeOut
  context.system.scheduler.schedule(0 millis, WORKER_TIMEOUT millis, self, CheckForWorkerTimeOut)

preStart方法里创建了一个定时器，定时检查Woker的超时时间 val WORKER_TIMEOUT = conf.getLong("spark.worker.timeout", 60) * 1000 默认为60秒

到此Master的初始化的主要过程到我们已经看到了，主要就是构造一个Master的Actor进行等待消息，并初始化了一堆集合来保存Worker信息，和一个定时器来检查Worker的超时

Master启动时序图

Woker的启动

通过Shell脚本执行salves.sh -> 通过读取slaves 通过ssh的方式启动远端的worker
spark-daemon.sh start org.apache.spark.deploy.worker.Worker

脚本会启动org.apache.spark.deploy.worker.Worker类

看Worker源码

private[spark] object Worker extends Logging {
  //Worker启动的入口
  def main(argStrings: Array[String]) {
    SignalLogger.register(log)
    val conf = new SparkConf
    val args = new WorkerArguments(argStrings, conf)
    //新创ActorSystem和Actor
    val (actorSystem, _) = startSystemAndActor(args.host, args.port, args.webUiPort, args.cores,
      args.memory, args.masters, args.workDir)
    actorSystem.awaitTermination()
  }

这里最重要的是Woker的startSystemAndActor

  def startSystemAndActor(
      host: String,
      port: Int,
      webUiPort: Int,
      cores: Int,
      memory: Int,
      masterUrls: Array[String],
      workDir: String,
      workerNumber: Option[Int] = None,
      conf: SparkConf = new SparkConf): (ActorSystem, Int) = {

    // The LocalSparkCluster runs multiple local sparkWorkerX actor systems
    val systemName = "sparkWorker" + workerNumber.map(_.toString).getOrElse("")
    val actorName = "Worker"
    val securityMgr = new SecurityManager(conf)
    //通过AkkaUtils ActorSystem
    val (actorSystem, boundPort) = AkkaUtils.createActorSystem(systemName, host, port,
      conf = conf, securityManager = securityMgr)
    val masterAkkaUrls = masterUrls.map(Master.toAkkaUrl(_, AkkaUtils.protocol(actorSystem)))
    //通过actorSystem.actorOf创建Actor   Worker-》执行构造器 -》 preStart -》 receice
    actorSystem.actorOf(Props(classOf[Worker], host, boundPort, webUiPort, cores, memory,
      masterAkkaUrls, systemName, actorName,  workDir, conf, securityMgr), name = actorName)
    (actorSystem, boundPort)
  }

这里Worker同样的构造了一个属于Worker的Actor对象,到此Worker的启动初始化完成

Worker与Master通信

根据Actor生命周期接着Worker的preStart方法被调用

  override def preStart() {
    assert(!registered)
    logInfo("Starting Spark worker %s:%d with %d cores, %s RAM".format(
      host, port, cores, Utils.megabytesToString(memory)))
    logInfo(s"Running Spark version ${org.apache.spark.SPARK_VERSION}")
    logInfo("Spark home: " + sparkHome)
    createWorkDir()
    context.system.eventStream.subscribe(self, classOf[RemotingLifecycleEvent])
    shuffleService.startIfEnabled()
    webUi = new WorkerWebUI(this, workDir, webUiPort)
    webUi.bind()

    //Worker向Master注册
    registerWithMaster()
    ....
  }

这里调用了一个registerWithMaster方法，开始向Master注册

 def registerWithMaster() {
    // DisassociatedEvent may be triggered multiple times, so don't attempt registration
    // if there are outstanding registration attempts scheduled.
    registrationRetryTimer match {
      case None =>
        registered = false
        //开始注册
        tryRegisterAllMasters()
        ....
    }
  }

registerWithMaster里通过匹配调用了tryRegisterAllMasters方法
，接下来看

  private def tryRegisterAllMasters() {
    //遍历master的地址
    for (masterAkkaUrl <- masterAkkaUrls) {
      logInfo("Connecting to master " + masterAkkaUrl + "...")
      //Worker跟Mater建立连接
      val actor = context.actorSelection(masterAkkaUrl)
      //向Master发送注册信息
      actor ! RegisterWorker(workerId, host, port, cores, memory, webUi.boundPort, publicAddress)
    }
  }

通过masterAkkaUrl和Master建立连接后
actor ! RegisterWorker(workerId, host, port, cores, memory, webUi.boundPort, publicAddress)Worker向Master发送了一个消息，带去一些参数，id,主机，端口，cpu核数，内存等待

override def receiveWithLogging = {
    ......
  
    //接受来自Worker的注册信息
    case RegisterWorker(id, workerHost, workerPort, cores, memory, workerUiPort, publicAddress) =>
    {
      logInfo("Registering worker %s:%d with %d cores, %s RAM".format(
        workerHost, workerPort, cores, Utils.megabytesToString(memory)))
      if (state == RecoveryState.STANDBY) {
        // ignore, don't send response
        //判断这个worker是否已经注册过
      } else if (idToWorker.contains(id)) {
        //如果注册过，告诉worker注册失败
        sender ! RegisterWorkerFailed("Duplicate worker ID")
      } else {
        //没有注册过，把来自Worker的注册信息封装到WorkerInfo当中
        val worker = new WorkerInfo(id, workerHost, workerPort, cores, memory,
          sender, workerUiPort, publicAddress)
        if (registerWorker(worker)) {
          //用持久化引擎记录Worker的信息
          persistenceEngine.addWorker(worker)
          //向Worker反馈信息，告诉Worker注册成功
          sender ! RegisteredWorker(masterUrl, masterWebUiUrl)
    
          schedule()
        } else {
          val workerAddress = worker.actor.path.address
          logWarning("Worker registration failed. Attempted to re-register worker at same " +
            "address: " + workerAddress)
          sender ! RegisterWorkerFailed("Attempted to re-register worker at same address: "
            + workerAddress)
        }
      }
    }

这里是最主要的内容；
receiveWithLogging里会轮询到Worker发送的消息，
Master收到消息后将参数封装成WorkInfo对象添加到集合中，并加入到持久化引擎中
sender ! RegisteredWorker(masterUrl, masterWebUiUrl)向Worker发送一个消息反馈

接下来看Worker的receiveWithLogging

override def receiveWithLogging = {
    
    case RegisteredWorker(masterUrl, masterWebUiUrl) =>
      logInfo("Successfully registered with master " + masterUrl)
      registered = true
      changeMaster(masterUrl, masterWebUiUrl)
      //启动定时器，定时发送心跳Heartbeat
      context.system.scheduler.schedule(0 millis, HEARTBEAT_MILLIS millis, self, SendHeartbeat)
      if (CLEANUP_ENABLED) {
        logInfo(s"Worker cleanup enabled; old application directories will be deleted in: $workDir")
        context.system.scheduler.schedule(CLEANUP_INTERVAL_MILLIS millis,
          CLEANUP_INTERVAL_MILLIS millis, self, WorkDirCleanup)
      }

worker接受来自Master的注册成功的反馈信息,启动定时器，定时发送心跳Heartbeat

    case SendHeartbeat =>
      //worker发送心跳的目的就是为了报活
      if (connected) { master ! Heartbeat(workerId) }

Master端的receiveWithLogging收到心跳消息

  override def receiveWithLogging = {
        ....
    case Heartbeat(workerId) => {
      idToWorker.get(workerId) match {
        case Some(workerInfo) =>
          //更新最后一次心跳时间
          workerInfo.lastHeartbeat = System.currentTimeMillis()
          .....
      }
    }
 }

记录并更新workerInfo.lastHeartbeat = System.currentTimeMillis()最后一次心跳时间

Master的定时任务会不断的发送一个CheckForWorkerTimeOut内部消息不断的轮询集合里的Worker信息，如果超过60秒就将Worker信息移除

  //检查超时的Worker
    case CheckForWorkerTimeOut => {
      timeOutDeadWorkers()
    }

timeOutDeadWorkers方法

  def timeOutDeadWorkers() {
    // Copy the workers into an array so we don't modify the hashset while iterating through it
    val currentTime = System.currentTimeMillis()
    val toRemove = workers.filter(_.lastHeartbeat < currentTime - WORKER_TIMEOUT).toArray
    for (worker <- toRemove) {
      if (worker.state != WorkerState.DEAD) {
        logWarning("Removing %s because we got no heartbeat in %d seconds".format(
          worker.id, WORKER_TIMEOUT/1000))
        removeWorker(worker)
      } else {
        if (worker.lastHeartbeat < currentTime - ((REAPER_ITERATIONS + 1) * WORKER_TIMEOUT)) {
          workers -= worker // we've seen this DEAD worker in the UI, etc. for long enough; cull it
        }
      }
    }
  }

如果 （最后一次心跳时间<当前时间-超时时间）则判断为Worker超时，
将集合里的信息移除。
当下一次收到心跳信息时，如果是已注册过的，workerId不为空，但是WorkerInfo已被移除的条件，就会sender ! ReconnectWorker(masterUrl)发送一个重新注册的消息

 case None =>
          if (workers.map(_.id).contains(workerId)) {
            logWarning(s"Got heartbeat from unregistered worker $workerId." +
              " Asking it to re-register.")
            //发送重新注册的消息
            sender ! ReconnectWorker(masterUrl)
          } else {
            logWarning(s"Got heartbeat from unregistered worker $workerId." +
              " This worker was never registered, so ignoring the heartbeat.")
          }

Worker与Master时序图