第10课：Spark Streaming源码解读之流数据不断接收全生命周期彻底研究和思考

写个大数据应用程序而言性能最重要，架构设计和稳定性

 

源码

ReceiverTracker.scala

start

/** Start the endpoint and receiver execution thread. */
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
  }
}

// This not being null means the tracker has been started and not stopped
private var endpoint: RpcEndpointRef = null

/** RpcEndpoint to receive messages from the receivers. */
private class ReceiverTrackerEndpoint(override val rpcEnv: RpcEnv) extends ThreadSafeRpcEndpoint {

launchReceivers

/**
 * Get the receivers from the ReceiverInputDStreams, distributes them to the
 * worker nodes as a parallel collection, and runs them.
 */
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))
}

receive

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)
    }
  case RestartReceiver(receiver) =>
    // Old scheduled executors minus the ones that are not active any more
    val oldScheduledExecutors = getStoredScheduledExecutors(receiver.streamId)
    val scheduledLocations = if (oldScheduledExecutors.nonEmpty) {
        // Try global scheduling again
        oldScheduledExecutors
      } else {
        val oldReceiverInfo = receiverTrackingInfos(receiver.streamId)
        // Clear "scheduledLocations" to indicate we are going to do local scheduling
        val newReceiverInfo = oldReceiverInfo.copy(
          state = ReceiverState.INACTIVE, scheduledLocations = None)
        receiverTrackingInfos(receiver.streamId) = newReceiverInfo
        schedulingPolicy.rescheduleReceiver(
          receiver.streamId,
          receiver.preferredLocation,
          receiverTrackingInfos,
          getExecutors)
      }
    // Assume there is one receiver restarting at one time, so we don't need to update
    // receiverTrackingInfos
    startReceiver(receiver, scheduledLocations)
  case c: CleanupOldBlocks =>
    receiverTrackingInfos.values.flatMap(_.endpoint).foreach(_.send(c))
  case UpdateReceiverRateLimit(streamUID, newRate) =>
    for (info <- receiverTrackingInfos.get(streamUID); eP <- info.endpoint) {
      eP.send(UpdateRateLimit(newRate))
    }
  // Remote messages
  case ReportError(streamId, message, error) =>
    reportError(streamId, message, error)
}

startReceiver

/**
 * Start a receiver along with its scheduled executors
 */
private def startReceiver(
    receiver: Receiver[_],
    scheduledLocations: Seq[TaskLocation]): Unit = {
  def shouldStartReceiver: Boolean = {
    // It's okay to start when trackerState is Initialized or Started
    !(isTrackerStopping || isTrackerStopped)
  }

  val receiverId = receiver.streamId
  if (!shouldStartReceiver) {
    onReceiverJobFinish(receiverId)
    return
  }

  val checkpointDirOption = Option(ssc.checkpointDir)
  val serializableHadoopConf =
    new SerializableConfiguration(ssc.sparkContext.hadoopConfiguration)

  // 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)
  logInfo(s"Receiver ${receiver.streamId} started")
}

ReceiverSupervisorImpl

//构造ReceiverSupervisorImpl

/** Remote RpcEndpointRef for the ReceiverTracker */
private val trackerEndpoint = RpcUtils.makeDriverRef("ReceiverTracker", env.conf, env.rpcEnv)

/** RpcEndpointRef for receiving messages from the ReceiverTracker in the driver */
private val endpoint = env.rpcEnv.setupEndpoint(
  "Receiver-" + streamId + "-" + System.currentTimeMillis(), new ThreadSafeRpcEndpoint {

ReceiverTracker就是消息通信体

endpoint = ssc.env.rpcEnv.setupEndpoint(
  "ReceiverTracker", new ReceiverTrackerEndpoint(ssc.env.rpcEnv))

构建了RPC消息通信体

/** RpcEndpointRef for receiving messages from the ReceiverTracker in the driver */
private val endpoint = env.rpcEnv.setupEndpoint(
  "Receiver-" + streamId + "-" + System.currentTimeMillis(), new ThreadSafeRpcEndpoint {

receive

override def receive: PartialFunction[Any, Unit] = {
  case StopReceiver =>
    logInfo("Received stop signal")
    ReceiverSupervisorImpl.this.stop("Stopped by driver", None)
  case CleanupOldBlocks(threshTime) =>
    logDebug("Received delete old batch signal")
    cleanupOldBlocks(threshTime)
  case UpdateRateLimit(eps) =>
    logInfo(s"Received a new rate limit: $eps.")
    registeredBlockGenerators.foreach { bg =>
      bg.updateRate(eps)
    }
}

UpdateRateLimit动态调整消息通信速度

private[receiver] def updateRate(newRate: Long): Unit =
  if (newRate > 0) {
    if (maxRateLimit > 0) {
      rateLimiter.setRate(newRate.min(maxRateLimit))
    } else {
      rateLimiter.setRate(newRate)
    }
  }