Akka源码分析-Remote-ActorSystem

　　前面的文章都是基于local模式分析的，现在我们简要分析一下在remote模式下，ActorSystem的创建过程。

final val ProviderClass: String =
      setup.get[BootstrapSetup]
        .flatMap(_.actorRefProvider).map(_.identifier)
        .getOrElse(getString("akka.actor.provider")) match {
          case "local"   ⇒ classOf[LocalActorRefProvider].getName
          // these two cannot be referenced by class as they may not be on the classpath
          case "remote"  ⇒ "akka.remote.RemoteActorRefProvider"
          case "cluster" ⇒ "akka.cluster.ClusterActorRefProvider"
          case fqcn      ⇒ fqcn
        }

　　之前我们分析过，在创建provider过程中，是通过ProviderClass来判断具体是哪种模式的。从ProviderClass源码来看，当我们配置akka.actor.provider为remote时，会创建akka.remote.RemoteActorRefProvider的实例。我们知道ActorSystem在start时候会去调用provider.init方法进行初始化。

def init(system: ActorSystemImpl): Unit = {
    local.init(system)

    actorRefResolveThreadLocalCache = ActorRefResolveThreadLocalCache(system)

    remotingTerminator = system.systemActorOf(
      remoteSettings.configureDispatcher(Props(classOf[RemotingTerminator], local.systemGuardian)),
      "remoting-terminator")

    val internals = Internals(
      remoteDaemon = {
        val d = new RemoteSystemDaemon(
          system,
          local.rootPath / "remote",
          rootGuardian,
          remotingTerminator,
          _log,
          untrustedMode = remoteSettings.UntrustedMode)
        local.registerExtraNames(Map(("remote", d)))
        d
      },
      transport =
        if (remoteSettings.Artery.Enabled) remoteSettings.Artery.Transport match {
          case ArterySettings.AeronUpd ⇒ new ArteryAeronUdpTransport(system, this)
          case ArterySettings.Tcp      ⇒ new ArteryTcpTransport(system, this, tlsEnabled = false)
          case ArterySettings.TlsTcp   ⇒ new ArteryTcpTransport(system, this, tlsEnabled = true)
        }
        else new Remoting(system, this))

    _internals = internals
    remotingTerminator ! internals

    _log = Logging.withMarker(eventStream, getClass.getName)

    // this enables reception of remote requests
    transport.start()

    _remoteWatcher = createRemoteWatcher(system)
    remoteDeploymentWatcher = createRemoteDeploymentWatcher(system)
  }

　　我们来结合RemoteActorRefProvider的构造函数和init函数来初步理解RemoteActorRefProvider的行为。首先在init方法的第一步就是调用local的init，通过local的类型我们发现这是一个LocalActorRefProvider，local的作用暂时不做分析，继续往下看。

　　下面创建了ActorRefResolveThreadLocalCache对象，从ActorRefResolveThreadLocalCache的定义来看（这里就不再贴出相关代码），它是一个ThreadLocal变量，且是一个实现了Lru的缓存器，缓存的内容是ActorRef，具体作用也忽略。remotingTerminator的具体作用也不做深入分析。

private final case class Internals(transport: RemoteTransport, remoteDaemon: InternalActorRef)
    extends NoSerializationVerificationNeeded

　　Internals的定义还是值得一看的，它有两个变量，其中transport的值应该是new Remoting(system, this)，remoteDaemon的值是RemoteSystemDaemon。然后调用了transport.start()，也就是Remoting的start。那么Remoting具体又是什呢？

// Start assumes that it cannot be followed by another start() without having a shutdown() first
  override def start(): Unit = {
    endpointManager match {
      case None ⇒
        log.info("Starting remoting")
        val manager: ActorRef = system.systemActorOf(
          configureDispatcher(Props(classOf[EndpointManager], provider.remoteSettings.config, log)).withDeploy(Deploy.local),
          Remoting.EndpointManagerName)
        endpointManager = Some(manager)

        try {
          val addressesPromise: Promise[Seq[(AkkaProtocolTransport, Address)]] = Promise()
          manager ! Listen(addressesPromise)

          val transports: Seq[(AkkaProtocolTransport, Address)] = Await.result(
            addressesPromise.future,
            StartupTimeout.duration)
          if (transports.isEmpty) throw new RemoteTransportException("No transport drivers were loaded.", null)

          transportMapping = transports.groupBy {
            case (transport, _) ⇒ transport.schemeIdentifier
          } map { case (k, v) ⇒ k → v.toSet }

          defaultAddress = transports.head._2
          addresses = transports.map { _._2 }.toSet

          log.info("Remoting started; listening on addresses :" + addresses.mkString("[", ", ", "]"))

          manager ! StartupFinished
          eventPublisher.notifyListeners(RemotingListenEvent(addresses))

        } catch {
          case e: TimeoutException ⇒
            notifyError("Startup timed out. This is usually related to actor system host setting or host name resolution misconfiguration.", e)
            throw e
          case NonFatal(e) ⇒
            notifyError("Startup failed", e)
            throw e
        }

      case Some(_) ⇒
        log.warning("Remoting was already started. Ignoring start attempt.")
    }
  }

　　在Remoting.start过程中，首先创建了EndpointManager，然后发送了一条Listen消息，并使用Await.result等待它的返回，然后又给EndpointManager发送了StartUpFinished。上面代码中的log.info("Remoting started; listening on addresses :" + addresses.mkString("[", ", ", "]"))还是值得我们关注的，毕竟我们启动remote模式的ActorSystem会经常看到这个日志信息。我们来看看EndpointManager收到Listen消息后做了哪些操作。

　　那么listens又是什么呢？

private def listens: Future[Seq[(AkkaProtocolTransport, Address, Promise[AssociationEventListener])]] = {
    /*
     * Constructs chains of adapters on top of each driver as given in configuration. The resulting structure looks
     * like the following:
     *   AkkaProtocolTransport <- Adapter <- ... <- Adapter <- Driver
     *
     * The transports variable contains only the heads of each chains (the AkkaProtocolTransport instances).
     */
    val transports: Seq[AkkaProtocolTransport] = for ((fqn, adapters, config) ← settings.Transports) yield {

      val args = Seq(classOf[ExtendedActorSystem] → context.system, classOf[Config] → config)

      // Loads the driver -- the bottom element of the chain.
      // The chain at this point:
      //   Driver
      val driver = extendedSystem.dynamicAccess
        .createInstanceFor[Transport](fqn, args).recover({

          case exception ⇒ throw new IllegalArgumentException(
            s"Cannot instantiate transport [$fqn]. " +
              "Make sure it extends [akka.remote.transport.Transport] and has constructor with " +
              "[akka.actor.ExtendedActorSystem] and [com.typesafe.config.Config] parameters", exception)

        }).get

      // Iteratively decorates the bottom level driver with a list of adapters.
      // The chain at this point:
      //   Adapter <- ... <- Adapter <- Driver
      val wrappedTransport =
        adapters.map { TransportAdaptersExtension.get(context.system).getAdapterProvider }.foldLeft(driver) {
          (t: Transport, provider: TransportAdapterProvider) ⇒
            // The TransportAdapterProvider will wrap the given Transport and returns with a wrapped one
            provider.create(t, context.system.asInstanceOf[ExtendedActorSystem])
        }

      // Apply AkkaProtocolTransport wrapper to the end of the chain
      // The chain at this point:
      //   AkkaProtocolTransport <- Adapter <- ... <- Adapter <- Driver
      new AkkaProtocolTransport(wrappedTransport, context.system, new AkkaProtocolSettings(conf), AkkaPduProtobufCodec)
    }

    // Collect all transports, listen addresses and listener promises in one future
    Future.sequence(transports.map { transport ⇒
      transport.listen map { case (address, listenerPromise) ⇒ (transport, address, listenerPromise) }
    })
  }

　　很明显这是一个transports集合，每个transports应该是一个AkkaProtocolTransport对象，AkkaProtocolTransport创建完成之后，调用了listen方法，最终返回AkkaProtocolTransport的列表。其实分析到这里我们可以不必再继续深入AkkaProtocolTransport的具体功能，从上面的官方注释以及我们的猜测来看，这大概是在初始化网络相关的对象。比如它可以是一个socket或者netty封装后的socket，是用来listen某个端口号，接收和发送数据的。

　　当然RemoteActorRefProvider.init的最后两行分别创建了RemoteWatcher、RemoteDeploymentWatcher，这两个Actor的作用后面再具体分析。

　　至此，remote模式下的初始化基本就算结束了，其实就是用RemoteActorRefProvider替换了LocalActorRefProvider，并完成了provider相关的初始化。remote模式与local模式下，ActorSystem初始化过程区别并不大，这还得多谢Akka框架封装的好。下一篇博客我们会分析actor的创建过程，毕竟在remote模式下，actor的创建过程还是有点不同的。