1.简述

spark 源码分析第一篇，准备从最基本的集群搭建入手，全面剖析spark。希望自己能对spark又更深入的理解。希望对读者有所帮助。言归正传，Spark standalone 模式，架构图：

这里先讨论Master和worker启动，以及之间的通讯：worker向master注册，worker向master发送heartbeat。

2.Master及启动流程

继承ThreadSafeRpcEndpoint类。启动master，会执行它自己的onStart函数。

2.1.执行start-master.sh脚本，->spark-daemon.sh "org.apache.spark.deploy.master.Master" -> spark-class start ->org.apache.spark.launcher.Main ->org.apache.spark.deploy.master.Master

2.2.调用Master的main方法，执行

val (rpcEnv, _, _) = startRpcEnvAndEndpoint(args.host, args.port, args.webUiPort, conf)

在此方法里创建RpcEnv对象和master对象：

def startRpcEnvAndEndpoint( host: String, port: Int, webUiPort: Int, conf: SparkConf): (RpcEnv, Int, Option[Int]) = { val securityMgr = new SecurityManager(conf) val rpcEnv = RpcEnv.create(SYSTEM_NAME, host, port, conf, securityMgr) val masterEndpoint = rpcEnv.setupEndpoint(ENDPOINT_NAME, new Master(rpcEnv, rpcEnv.address, webUiPort, securityMgr, conf)) val portsResponse = masterEndpoint.askSync[BoundPortsResponse](BoundPortsRequest) (rpcEnv, portsResponse.webUIPort, portsResponse.restPort) }

2.3.创建master对象，执行onstart方法。

(1).onstart函数做了如下的事情：

启动web UI

发送CheckForWorkerTimeOut消息给自己，移除超时的worker。

根据参数判断是否启动rest的接口，

注册master resources到master MetricsSystem

根据参数指定的recovery mode进行恢复。

(2) receive 函数接收了哪些消息

ElectedLeader

CompleteRecovery

RevokedLeadership 收回leader权力

RegisterWorker 注册worker ：注册成功后，发送RegisteredWorker消息给worker

RegisterApplication 注册application

ExecutorStateChanged

DriverStateChanged

Heartbeat

MasterChangeAcknowledged

WorkerSchedulerStateResponse

WorkerLatestState

CheckForWorkerTimeOut

RequestSubmitDriver 请求提交driver

RequestKillDriver

RequestDriverStatus

RequestMasterState

BoundPortsRequest

3.worker及启动流程

同样继承ThreadSafeRpcEndpoint类.

3.1.执行start-slave.sh脚本 -> spark-daemon.sh start "org.apache.spark.deploy.worker.Worker" ->-> spark-class start ->org.apache.spark.launcher.Main ->org.apache.spark.deploy.worker.Worker

3.2.执行Worker的main方法

启动rpcEnv对象和Worker对象。

val rpcEnv = startRpcEnvAndEndpoint(args.host, args.port, args.webUiPort, args.cores, args.memory, args.masters, args.workDir, conf = conf)

def startRpcEnvAndEndpoint( host: String, port: Int, webUiPort: Int, cores: Int, memory: Int, masterUrls: Array[String], workDir: String, workerNumber: Option[Int] = None, conf: SparkConf = new SparkConf): RpcEnv = {  // The LocalSparkCluster runs multiple local sparkWorkerX RPC Environments val systemName = SYSTEM_NAME + workerNumber.map(_.toString).getOrElse("") val securityMgr = new SecurityManager(conf) val rpcEnv = RpcEnv.create(systemName, host, port, conf, securityMgr) val masterAddresses = masterUrls.map(RpcAddress.fromSparkURL(_)) rpcEnv.setupEndpoint(ENDPOINT_NAME, new Worker(rpcEnv, webUiPort, cores, memory, masterAddresses, ENDPOINT_NAME, workDir, conf, securityMgr)) rpcEnv }

3.3.创建worker对象，执行onstart方法。

(1).onstart函数做了如下的事情：

在SPAKR_HOME目录下创建work目录

启动外部shuffle服务

启动workerweb UI

注册到master上，根据 val masterEndpoint = rpcEnv.setupEndpointRef(masterAddress, Master.ENDPOINT_NAME)获取masterRef。

启动MetricsSystem

3.4.接收到master注册完成的消息RegisteredWorker之后调用handleRegisterResponse方法

private def handleRegisterResponse(msg: RegisterWorkerResponse): Unit = synchronized { msg match { case RegisteredWorker(masterRef, masterWebUiUrl, masterAddress) => if (preferConfiguredMasterAddress) { logInfo("Successfully registered with master " + masterAddress.toSparkURL) } else { logInfo("Successfully registered with master " + masterRef.address.toSparkURL) } registered = true //设置masterRef changeMaster(masterRef, masterWebUiUrl, masterAddress) //定时发送heartbeat forwordMessageScheduler.scheduleAtFixedRate(new Runnable { override def run(): Unit = Utils.tryLogNonFatalError { self.send(SendHeartbeat) } }, 0, HEARTBEAT_MILLIS, TimeUnit.MILLISECONDS) //是否清理workdir if (CLEANUP_ENABLED) { logInfo( s"Worker cleanup enabled; old application directories will be deleted in: $workDir") forwordMessageScheduler.scheduleAtFixedRate(new Runnable { override def run(): Unit = Utils.tryLogNonFatalError { self.send(WorkDirCleanup) } }, CLEANUP_INTERVAL_MILLIS, CLEANUP_INTERVAL_MILLIS, TimeUnit.MILLISECONDS) }  val execs = executors.values.map { e => new ExecutorDescription(e.appId, e.execId, e.cores, e.state) } masterRef.send(WorkerLatestState(workerId, execs.toList, drivers.keys.toSeq))  case RegisterWorkerFailed(message) => if (!registered) { logError("Worker registration failed: " + message) System.exit(1) }  case MasterInStandby => // Ignore. Master not yet ready. } }

4.总结

最好是跟着上面的讲接把源码理一遍，就清晰很多了， 接下来会分析driver的启动。