贯通Spark Streaming流计算框架的运行源码(第五篇)

本期内容：
基于简单案例贯通Spark Streaming的运行源码

我们下面是依然一个单词计数的简单案例，来开启逐步揭秘Spark Streaming的运行源码之旅

//新浪微博：http://weibo.com/ilovepains/

SparkConf conf = new SparkConf().setMaster("spark://Master:7077").setAppName("WordCountOnline");
        JavaStreamingContext jsc = new JavaStreamingContext(conf, Durations.seconds(5));
        JavaReceiverInputDStream lines = jsc.socketTextStream("Master", 9999);

        JavaDStream<String> words = lines.flatMap(new FlatMapFunction<String, String>() {
            @Override
            public Iterable<String> call(String line) throws Exception {
                return Arrays.asList(line.split(" "));
            }
        });

        JavaPairDStream<String, Integer> pairs = words.mapToPair(new PairFunction<String, String, Integer>() {

            @Override
            public Tuple2<String, Integer> call(String word) throws Exception {
                return new Tuple2<String, Integer>(word, 1);
            }
        });

        JavaPairDStream<String, Integer> wordsCount = pairs.reduceByKey(new Function2<Integer, Integer, Integer>() { 

            @Override
            public Integer call(Integer v1, Integer v2) throws Exception {
                return v1 + v2;
            }
        });

        wordsCount.print();

        jsc.start();
        jsc.awaitTermination();
        jsc.close();

1、我们从构建StreamingContext开始，如这句代码：
JavaStreamingContext jsc = new JavaStreamingContext(conf, Durations.seconds(5))，我们来看其背后的源码：

//StreamingContext.scala 80行
  def this(conf: SparkConf, batchDuration: Duration) = {
    this(StreamingContext.createNewSparkContext(conf), null, batchDuration)
  }
//StreamingContext.scala 873行
private[streaming] def createNewSparkContext(conf: SparkConf): SparkContext = {
    new SparkContext(conf)
  }

由上源码我们可以看到，在构建St传入的reamingContext时，除了SparkConf参数（里面含有Configuration信息）和时间间隔batchDuration参数，其实在内部构建了SparkContext，这也意味着StreamingContext依赖于SparkContext，也印证了我们前面文档的一句话：SparkStreaming是运行于SparkCode之上的一个子框架

2、我们接着来看这句代码背后的机密:JavaReceiverInputDStream lines = jsc.socketTextStream(“Master”, 9999);

//StreamingContext 327行
def socketTextStream( hostname: String, port: Int, storageLevel: StorageLevel = StorageLevel.MEMORY_AND_DISK_SER_2 ): ReceiverInputDStream[String] = withNamedScope("socket text stream") {
    socketStream[String](hostname, port, SocketReceiver.bytesToLines, storageLevel)
  }

//StreamingContext 345行
def socketStream[T: ClassTag](
      hostname: String,
      port: Int,
      converter: (InputStream) => Iterator[T],
      storageLevel: StorageLevel
    ): ReceiverInputDStream[T] = {
    new SocketInputDStream[T](this, hostname, port, converter, storageLevel)
  }

我们可以在看到在调用socketTextStream时返回的是ReceiverInputDStream对象，并且我们可以看到socket输入流接受数据默认的storageLevel是MEMORY_AND_DISK_SER_2

2.1、创建SocketInputDStream输入流(SocketInputDStream继承于ReceiverInputDStream，而ReceiverInputDStream继承于InputDStream，InputDStream继承于DStream)，通过构建SocketReceiver来接受数据

//SocketInputDStream.scala 33行
private[streaming]
class SocketInputDStream[T: ClassTag](
    ssc_ : StreamingContext,
    host: String,
    port: Int,
    bytesToObjects: InputStream => Iterator[T],
    storageLevel: StorageLevel
  ) extends ReceiverInputDStream[T](ssc_) {

  def getReceiver(): Receiver[T] = {
    new SocketReceiver(host, port, bytesToObjects, storageLevel)
  }
}

2.2、而SocketReceiver是通过其本身onStart方法开启接受数据

//SocketInputDStream.scala 47行
def onStart() {
    // Start the thread that receives data over a connection
    new Thread("Socket Receiver") {
      setDaemon(true)
      override def run() { receive() }
    }.start()
  }

2.3、这时候我们发现onStart方法又调用了receive方法

//SocketInputDStream.scala 69行
def receive() {
    var socket: Socket = null
    try {
      logInfo("Connecting to " + host + ":" + port)
      **socket = new Socket(host, port)**
      logInfo("Connected to " + host + ":" + port)
      val iterator = bytesToObjects(socket.getInputStream())
      while(!isStopped && iterator.hasNext) {
        store(iterator.next)
      }
      if (!isStopped()) {
        restart("Socket data stream had no more data")
      } else {
        logInfo("Stopped receiving")
      }
    } catch {
      case e: java.net.ConnectException =>
        restart("Error connecting to " + host + ":" + port, e)
      case NonFatal(e) =>
        logWarning("Error receiving data", e)
        restart("Error receiving data", e)
    } finally {
      if (socket != null) {
        socket.close()
        logInfo("Closed socket to " + host + ":" + port)
      }
    }
  }

至此我们看到receive方法中实际上是通过Socket网络通信的方式来接受数据的

3、 我们接着来看jsc.start()背后源码：

//StreamingContext.scala 606行
  // Start the streaming scheduler in a new thread, so that thread local properties
            // like call sites and job groups can be reset without affecting those of the
            // current thread.
            ThreadUtils.runInNewThread("streaming-start") {
              sparkContext.setCallSite(startSite.get)
              sparkContext.clearJobGroup()
              sparkContext.setLocalProperty(SparkContext.SPARK_JOB_INTERRUPT_ON_CANCEL, "false")
              scheduler.start()
            }

我们可以看到在StreamingContext调用start方法，而start里面是通过线程池的开辟一条线程去执行JobScheduler.start()启动，并且该线程设置了一些私有属性以便于于其他之间线程互不影响

3.1、在JobScheduler.start()方法中，创建EventLoop消息线程并执行start

//JobScheduler.scala 66行
 eventLoop = new EventLoop[JobSchedulerEvent]("JobScheduler") {
      override protected def onReceive(event: JobSchedulerEvent): Unit = processEvent(event)

      override protected def onError(e: Throwable): Unit = reportError("Error in job scheduler", e)
    }
    eventLoop.start()

3.2 在eventLoop.start()方法的背后，实际上最终还是调用是processEvent(event)方法

//EventLoop.scala 40行
private val eventThread = new Thread(name) {
    setDaemon(true)

    override def run(): Unit = {
      try {
        while (!stopped.get) {
          val event = eventQueue.take()
          try {
            onReceive(event)
          } catch {
            case NonFatal(e) => {
              try {
                onError(e)
              } catch {
                case NonFatal(e) => logError("Unexpected error in " + name, e)
              }
            }
          }
        }
      } catch {
        case ie: InterruptedException => // exit even if eventQueue is not empty
        case NonFatal(e) => logError("Unexpected error in " + name, e)
      }
    }

  }

3.3、JobSchedule是Job的核心调度器，通过EventLoop的processEvent方法循环监听各个Job的执行状态(包括Job的启动、完成、报错)

private def processEvent(event: JobSchedulerEvent) {
    try {
      event match {
        case JobStarted(job, startTime) => handleJobStart(job, startTime)
        case JobCompleted(job, completedTime) => handleJobCompletion(job, completedTime)
        case ErrorReported(m, e) => handleError(m, e)
      }
    } catch {
      case e: Throwable =>
        reportError("Error in job scheduler", e)
    }
  }

3.4、我们接着回到JobScheduler.start，接着往下看

 // attach rate controllers of input streams to receive batch completion updates
    for {
      inputDStream <- ssc.graph.getInputStreams
      rateController <- inputDStream.rateController
    } ssc.addStreamingListener(rateController)

从注释我们看到这块代码主要用于监听输入流的状态并可以控制输入流的流速
3.5、接着往下走：StreamingListenerBus

//StreamingListenerBus.scala 32行
  listenerBus.start(ssc.sparkContext)
    receiverTracker = new ReceiverTracker(ssc)
    inputInfoTracker = new InputInfoTracker(ssc)
    receiverTracker.start()
    jobGenerator.start()

上面可以看到调用启动了StreamingListenerBus线程，StreamingListenerBus最终调用onPostEvent方法

override def onPostEvent(listener: StreamingListener, event: StreamingListenerEvent): Unit = {
    event match {
      case receiverStarted: StreamingListenerReceiverStarted =>
        listener.onReceiverStarted(receiverStarted)
      case receiverError: StreamingListenerReceiverError =>
        listener.onReceiverError(receiverError)
      case receiverStopped: StreamingListenerReceiverStopped =>
        listener.onReceiverStopped(receiverStopped)
      case batchSubmitted: StreamingListenerBatchSubmitted =>
        listener.onBatchSubmitted(batchSubmitted)
      case batchStarted: StreamingListenerBatchStarted =>
        listener.onBatchStarted(batchStarted)
      case batchCompleted: StreamingListenerBatchCompleted =>
        listener.onBatchCompleted(batchCompleted)
      case outputOperationStarted: StreamingListenerOutputOperationStarted =>
        listener.onOutputOperationStarted(outputOperationStarted)
      case outputOperationCompleted: StreamingListenerOutputOperationCompleted =>
        listener.onOutputOperationCompleted(outputOperationCompleted)
      case _ =>
    }
  }

从StreamingListenerBus下面注释我们可以看出StreamingListenerBus主要用于提交SparkStreaming中各种事件至StreamingListeners监听执行情况

/** Asynchronously passes StreamingListenerEvents to registered StreamingListeners. */
private[spark] class StreamingListenerBus
  extends AsynchronousListenerBus[StreamingListener, StreamingListenerEvent]("StreamingListenerBus")
  with Logging {

3.6、继续回到JonSchedule.start中的receiverTracker.start()

//ReceiverTracker.scala 149行
def start(): Unit = synchronized {
    if (isTrackerStarted) {
      throw new SparkException("ReceiverTracker already started")
    }

    if (!receiverInputStreams.isEmpty) {
      endpoint = ssc.env.rpcEnv.setupEndpoint(
        "ReceiverTracker", new ReceiverTrackerEndpoint(ssc.env.rpcEnv))
      if (!skipReceiverLaunch) launchReceivers()
      logInfo("ReceiverTracker started")
      trackerState = Started
    }
  }

3.7 我们重点关注一下：ReceiverTracker.launchReceivers

//ReceiverTracker.scala 413行
private def launchReceivers(): Unit = {
    val receivers = receiverInputStreams.map(nis => {
      val rcvr = nis.getReceiver()
      rcvr.setReceiverId(nis.id)
      rcvr
    })

    runDummySparkJob()

    logInfo("Starting " + receivers.length + " receivers")
    endpoint.send(StartAllReceivers(receivers))
  }

3.8 runDummySparkJob：运行了一个Dummy的作业，确保所有的Slaves正常工作，避免所有的Receiver都在一台机器上

/**
   * Run the dummy Spark job to ensure that all slaves have registered. This avoids all the
   * receivers to be scheduled on the same node.
   *
   * TODO Should poll the executor number and wait for executors according to
   * "spark.scheduler.minRegisteredResourcesRatio" and
   * "spark.scheduler.maxRegisteredResourcesWaitingTime" rather than running a dummy job.
   */
  private def runDummySparkJob(): Unit = {
    if (!ssc.sparkContext.isLocal) {
      ssc.sparkContext.makeRDD(1 to 50, 50).map(x => (x, 1)).reduceByKey(_ + _, 20).collect()
    }
    assert(getExecutors.nonEmpty)
  }

这段代码也解释了为什么我们在第一篇博客中出现Job0的情况
3.9 endpoint.send(StartAllReceivers(receivers)

//ReceiverTracker.scala 155行
 endpoint = ssc.env.rpcEnv.setupEndpoint(
        "ReceiverTracker", new ReceiverTrackerEndpoint(ssc.env.rpcEnv))

3.9.1、由上面源码可知endpoint指向的是ReceiverTrackerEndpoint，我们再关注一下ReceiverTrackerEndpoint.receive

//ReceiverTracker.scala 447行
override def receive: PartialFunction[Any, Unit] = {
      // Local messages
      case StartAllReceivers(receivers) =>
        val scheduledLocations = schedulingPolicy.scheduleReceivers(receivers, getExecutors)
        for (receiver <- receivers) {
          val executors = scheduledLocations(receiver.streamId)
          updateReceiverScheduledExecutors(receiver.streamId, executors)
          receiverPreferredLocations(receiver.streamId) = receiver.preferredLocation
          startReceiver(receiver, executors)
        }

3.9.2、startReceiver(receiver, executors)，startReceiver会在worker节点中启动一个Job用于不断的接受数据并转换成对应的RDD

//ReceiverTracker.scala 563行
 // Function to start the receiver on the worker node
      val startReceiverFunc: Iterator[Receiver[_]] => Unit =
        (iterator: Iterator[Receiver[_]]) => {
          if (!iterator.hasNext) {
            throw new SparkException(
              "Could not start receiver as object not found.")
          }
          if (TaskContext.get().attemptNumber() == 0) {
            val receiver = iterator.next()
            assert(iterator.hasNext == false)
            val supervisor = new ReceiverSupervisorImpl(
              receiver, SparkEnv.get, serializableHadoopConf.value, checkpointDirOption)
            supervisor.start()
            supervisor.awaitTermination()
          } else {
            // It's restarted by TaskScheduler, but we want to reschedule it again. So exit it.
          }
        }

      // Create the RDD using the scheduledLocations to run the receiver in a Spark job
      val receiverRDD: RDD[Receiver[_]] =
        if (scheduledLocations.isEmpty) {
          ssc.sc.makeRDD(Seq(receiver), 1)
        } else {
          val preferredLocations = scheduledLocations.map(_.toString).distinct
          ssc.sc.makeRDD(Seq(receiver -> preferredLocations))
        }
      receiverRDD.setName(s"Receiver $receiverId")
      ssc.sparkContext.setJobDescription(s"Streaming job running receiver $receiverId")
      ssc.sparkContext.setCallSite(Option(ssc.getStartSite()).getOrElse(Utils.getCallSite()))

      val future = ssc.sparkContext.submitJob[Receiver[_], Unit, Unit](
        receiverRDD, startReceiverFunc, Seq(0), (_, _) => Unit, ())
      // We will keep restarting the receiver job until ReceiverTracker is stopped
      future.onComplete {
        case Success(_) =>
          if (!shouldStartReceiver) {
            onReceiverJobFinish(receiverId)
          } else {
            logInfo(s"Restarting Receiver $receiverId")
            self.send(RestartReceiver(receiver))
          }
        case Failure(e) =>
          if (!shouldStartReceiver) {
            onReceiverJobFinish(receiverId)
          } else {
            logError("Receiver has been stopped. Try to restart it.", e)
            logInfo(s"Restarting Receiver $receiverId")
            self.send(RestartReceiver(receiver))
          }
      }(submitJobThreadPool)

4、回到JobScheduler.start 的 jobGenerator.start()方法

//JobGenerator.scala 79行
def start(): Unit = synchronized {
    if (eventLoop != null) return // generator has already been started

    // Call checkpointWriter here to initialize it before eventLoop uses it to avoid a deadlock.
    // See SPARK-10125
    checkpointWriter

    eventLoop = new EventLoop[JobGeneratorEvent]("JobGenerator") {
      override protected def onReceive(event: JobGeneratorEvent): Unit = processEvent(event)

      override protected def onError(e: Throwable): Unit = {
        jobScheduler.reportError("Error in job generator", e)
      }
    }
    eventLoop.start()

    if (ssc.isCheckpointPresent) {
      restart()
    } else {
      startFirstTime()
    }
  }

此处我们看到了EventLoop线程，我们进入EventLoop最终调用的processEvent方法

//JobGenerator.scala 178行
private def processEvent(event: JobGeneratorEvent) {
    logDebug("Got event " + event)
    event match {
      case GenerateJobs(time) => generateJobs(time)
      case ClearMetadata(time) => clearMetadata(time)
      case DoCheckpoint(time, clearCheckpointDataLater) =>
        doCheckpoint(time, clearCheckpointDataLater)
      case ClearCheckpointData(time) => clearCheckpointData(time)
    }
  }

generateJobs(time)方法

/** Generate jobs and perform checkpoint for the given `time`. */
  private def generateJobs(time: Time) {
    // Set the SparkEnv in this thread, so that job generation code can access the environment // Example: BlockRDDs are created in this thread, and it needs to access BlockManager // Update: This is probably redundant after threadlocal stuff in SparkEnv has been removed. SparkEnv.set(ssc.env) Try { jobScheduler.receiverTracker.allocateBlocksToBatch(time) // allocate received blocks to batch graph.generateJobs(time) // generate jobs using allocated block } match { case Success(jobs) => val streamIdToInputInfos = jobScheduler.inputInfoTracker.getInfo(time) jobScheduler.submitJobSet(JobSet(time, jobs, streamIdToInputInfos)) case Failure(e) => jobScheduler.reportError("Error generating jobs for time " + time, e) } eventLoop.post(DoCheckpoint(time, clearCheckpointDataLater = false)) }

如下JobScheduler的submitJobSet方法，向线程池中提交JobHandler

def submitJobSet(jobSet: JobSet) {
    if (jobSet.jobs.isEmpty) {
      logInfo("No jobs added for time " + jobSet.time)
    } else {
      listenerBus.post(StreamingListenerBatchSubmitted(jobSet.toBatchInfo))
      jobSets.put(jobSet.time, jobSet)
      jobSet.jobs.foreach(job => jobExecutor.execute(new JobHandler(job)))
      logInfo("Added jobs for time " + jobSet.time)
    }
  }

JobHandler实现了Runnable 接口，最终调用了job.run()这个方法

private class JobHandler(job: Job) extends Runnable with Logging {
    import JobScheduler._

    def run() {
      try {
        val formattedTime = UIUtils.formatBatchTime(
          job.time.milliseconds, ssc.graph.batchDuration.milliseconds, showYYYYMMSS = false)
        val batchUrl = s"/streaming/batch/?id=${job.time.milliseconds}"
        val batchLinkText = s"[output operation ${job.outputOpId}, batch time ${formattedTime}]"

        ssc.sc.setJobDescription(
          s"""Streaming job from <a href="$batchUrl">$batchLinkText</a>""")
        ssc.sc.setLocalProperty(BATCH_TIME_PROPERTY_KEY, job.time.milliseconds.toString)
        ssc.sc.setLocalProperty(OUTPUT_OP_ID_PROPERTY_KEY, job.outputOpId.toString)

        // We need to assign `eventLoop` to a temp variable. Otherwise, because
        // `JobScheduler.stop(false)` may set `eventLoop` to null when this method is running, then
        // it's possible that when `post` is called, `eventLoop` happens to null.
        var _eventLoop = eventLoop
        if (_eventLoop != null) {
          _eventLoop.post(JobStarted(job, clock.getTimeMillis()))
          // Disable checks for existing output directories in jobs launched by the streaming
          // scheduler, since we may need to write output to an existing directory during checkpoint
          // recovery; see SPARK-4835 for more details.
          PairRDDFunctions.disableOutputSpecValidation.withValue(true) {
            job.run()
          }
          _eventLoop = eventLoop
          if (_eventLoop != null) {
            _eventLoop.post(JobCompleted(job, clock.getTimeMillis()))
          }
        } else {
          // JobScheduler has been stopped.
        }
      } finally {
        ssc.sc.setLocalProperty(JobScheduler.BATCH_TIME_PROPERTY_KEY, null)
        ssc.sc.setLocalProperty(JobScheduler.OUTPUT_OP_ID_PROPERTY_KEY, null)
      }
    }
  }

我们简单总结一下：
在StreamingContext调用start方法的内部其实是会启动JobScheduler的Start方法，进行消息循环，在JobScheduler的start内部会构造JobGenerator和ReceiverTacker，并且调用JobGenerator和ReceiverTacker的start方法：
1，JobGenerator启动后会不断的根据batchDuration生成一个个的Job
2，ReceiverTracker启动后首先在Spark Cluster中启动Receiver（其实是在Executor中先启动ReceiverSupervisor），在Receiver收到数据后会通过ReceiverSupervisor存储到Executor并且把数据的Metadata信息发送给Driver中的ReceiverTracker，在ReceiverTracker内部会通过ReceivedBlockTracker来管理接受到的元数据信息每个BatchInterval会产生一个具体的Job，其实这里的Job不是Spark Core中所指的Job，它只是基于DStreamGraph而生成的RDD的DAG而已，从Java角度讲，相当于Runnable接口实例，此时要想运行Job需要提交给JobScheduler，在JobScheduler中通过线程池的方式找到一个单独的线程来提交Job到集群运行（其实是在线程中基于RDD的Action触发真正的作业的运行），为什么使用线程池呢？
1，作业不断生成，所以为了提升效率，我们需要线程池；这和在Executor中通过线程池执行Task有异曲同工之妙；
2，有可能设置了Job的FAIR公平调度的方式，这个时候也需要多线程的支持；