sparkLauncher源码之LauncherServer

1.这个类主要作用是接受消息然后调用listener监听


 *
 *         -----------------------                       -----------------------
 *         |      User App       |     spark-submit      |      Spark App      |
 *         |                     |  -------------------> |                     |
 *         |         ------------|                       |-------------        |
 *         |         |           |        hello          |            |        |
 *         |         | L. Server |<----------------------| L. Backend |        |
 *         |         |           |                       |            |        |
 *         |         -------------                       -----------------------
 *         |               |     |                              ^
 *         |               v     |                              |
 *         |        -------------|                              |
 *         |        |            |      <per-app channel>       |
 *         |        | App Handle |<------------------------------
 *         |        |            |
 *         -----------------------
1.LauncherServer主要有两个作用:一个作用是可以定时关闭client连接的线程，一个作用是创建ChildProcAppHandle.
2.LauncherConnection  这个类是抽离出来的一个连接类，重要方法handle(Message meg)

3.LauncherBackend 主要用来与服务端通信class LauncherServer implements Closeable {
  private static final Logger LOG = Logger.getLogger(LauncherServer.class.getName());
  private static final String THREAD_NAME_FMT = "LauncherServer-%d";
//连接超时
  private static final long DEFAULT_CONNECT_TIMEOUT = 10000L;

//与子进程通信的密钥
 private static final SecureRandom RND = new SecureRandom();

  private static volatile LauncherServer serverInstance;


  static synchronized ChildProcAppHandle newAppHandle() throws IOException {
//初始化LauncherServer
 LauncherServer server = serverInstance != null ? serverInstance : new LauncherServer();    //标记数量
    server.ref();
//该对象持有这个server
    serverInstance = server;
//创建密钥
    String secret = server.createSecret();
//已经有了重新创建
    while (server.pending.containsKey(secret)) {
      secret = server.createSecret();
    }
//返回server的AppHandle
    return server.newAppHandle(secret);
  }

  static LauncherServer getServerInstance() {
    return serverInstance;
  }

  private final AtomicLong refCount;
  private final AtomicLong threadIds;
//标记服务器有的Handle
  private final ConcurrentMap<String, ChildProcAppHandle> pending;
//连接的客户端
  private final List<ServerConnection> clients;
  private final ServerSocket server;
  private final Thread serverThread;
  private final ThreadFactory factory;
  private final Timer timeoutTimer;

  private volatile boolean running;

  private LauncherServer() throws IOException {
//初始化个数
    this.refCount = new AtomicLong(0);
//初始化ServerSocket
    ServerSocket server = new ServerSocket();
    try {
//端口复用
      server.setReuseAddress(true);
      server.bind(new InetSocketAddress(InetAddress.getLoopbackAddress(), 0));

      this.clients = new ArrayList<>();
      this.threadIds = new AtomicLong();
      this.factory = new NamedThreadFactory(THREAD_NAME_FMT);
      this.pending = new ConcurrentHashMap<>();
//Timer类【实际上是一个线程套一个无限循环】的定时器。
      this.timeoutTimer = new Timer("LauncherServer-TimeoutTimer", true);
      this.server = server;
      this.running = true;
//服务器接受连接
      this.serverThread = factory.newThread(new Runnable() {
        @Override
        public void run() {
          acceptConnections();
        }
      });
//服务器启动
      serverThread.start();
    } catch (IOException ioe) {
      close();
      throw ioe;
    } catch (Exception e) {
      close();
      throw new IOException(e);
    }
  }


  ChildProcAppHandle newAppHandle(String secret) {
//注册child handle
    ChildProcAppHandle handle = new ChildProcAppHandle(secret, this);
//重复的key不会覆盖
    ChildProcAppHandle existing = pending.putIfAbsent(secret, handle);
    CommandBuilderUtils.checkState(existing == null, "Multiple handles with the same secret.");
    return handle;
  }
//以下主要是连接关闭的清除操作
  @Override
  public void close() throws IOException {
    synchronized (this) {//对象锁
      if (running) {
        running = false;
        timeoutTimer.cancel();//清除执行队列
        server.close();
        synchronized (clients) {
          List<ServerConnection> copy = new ArrayList<>(clients);
          clients.clear();
          for (ServerConnection client : copy) {
            client.close();
          }
        }
      }
    }
    if (serverThread != null) {
      try {
        serverThread.join();
      } catch (InterruptedException ie) {
        // no-op
      }
    }
  }
//累加计数
  void ref() {
    refCount.incrementAndGet();
  }

  void unref() {
    synchronized(LauncherServer.class) {
      if (refCount.decrementAndGet() == 0) {
        try {
          close();
        } catch (IOException ioe) {
          // no-op.
        } finally {
          serverInstance = null;
        }
      }
    }
  }
//服务器端口
  int getPort() {
    return server.getLocalPort();
  }


  void unregister(ChildProcAppHandle handle) {
    pending.remove(handle.getSecret());
    unref();
  }

  private void acceptConnections() {
    try {
      while (running) {
//服务器等待
        final Socket client = server.accept();
//定义一个线程关闭的操作
        TimerTask timeout = new TimerTask() {
          @Override
          public void run() {
            LOG.warning("Timed out waiting for hello message from client.");
            try {
              client.close();
            } catch (IOException ioe) {
              // no-op.
            }
          }
        };
//涉及到了LauncherConnection初始化了输出对象        ServerConnection clientConnection = new ServerConnection(client, timeout);        Thread clientThread = factory.newThread(clientConnection);
        synchronized (timeout) {
          clientThread.start();
          synchronized (clients) {
//建立连接并通信，添加客户端到列表中
            clients.add(clientConnection);
          }
//得到超时时间 spark.launcher.childConectionTimeout 默认10000L
         long timeoutMs = getConnectionTimeout();          // 0 is used for testing to avoid issues with clock resolution / thread scheduling,
          // and force an immediate timeout.
          if (timeoutMs > 0) {
//到一定时间就关掉连接
            timeoutTimer.schedule(timeout, getConnectionTimeout());
          } else {
            timeout.run();
          }
        }
      }
    } catch (IOException ioe) {
      if (running) {
        LOG.log(Level.SEVERE, "Error in accept loop.", ioe);
      }
    }
  }

//超时配置  
private long getConnectionTimeout() {
    String value = SparkLauncher.launcherConfig.get(SparkLauncher.CHILD_CONNECTION_TIMEOUT);
    return (value != null) ? Long.parseLong(value) : DEFAULT_CONNECT_TIMEOUT;
  }
//加密算法
  private String createSecret() {
    byte[] secret = new byte[128];
    RND.nextBytes(secret);

    StringBuilder sb = new StringBuilder();
    for (byte b : secret) {
      int ival = b >= 0 ? b : Byte.MAX_VALUE - b;
      if (ival < 0x10) {
        sb.append("0");
      }
      sb.append(Integer.toHexString(ival));
    }
    return sb.toString();
  }

  private class ServerConnection extends LauncherConnection {

    private TimerTask timeout;
    private ChildProcAppHandle handle;

    ServerConnection(Socket socket, TimerTask timeout) throws IOException {
      super(socket);
      this.timeout = timeout;
    }
//多种消息类型读取识别，并且会调用用户提供的listener
    @Override
    protected void handle(Message msg) throws IOException {
      try {
        if (msg instanceof Hello) {
          timeout.cancel();
          timeout = null;
          Hello hello = (Hello) msg;
          ChildProcAppHandle handle = pending.remove(hello.secret);
          if (handle != null) {
            handle.setConnection(this);
            handle.setState(SparkAppHandle.State.CONNECTED);
            this.handle = handle;
          } else {
            throw new IllegalArgumentException("Received Hello for unknown client.");
          }
        } else {
          if (handle == null) {
            throw new IllegalArgumentException("Expected hello, got: " +
            msg != null ? msg.getClass().getName() : null);
          }
          if (msg instanceof SetAppId) {
            SetAppId set = (SetAppId) msg;
            handle.setAppId(set.appId);
          } else if (msg instanceof SetState) {
            handle.setState(((SetState)msg).state);
          } else {
            throw new IllegalArgumentException("Invalid message: " +
              msg != null ? msg.getClass().getName() : null);
          }
        }
      } catch (Exception e) {
        LOG.log(Level.INFO, "Error handling message from client.", e);
        if (timeout != null) {
          timeout.cancel();
        }
        close();
      } finally {
        timeoutTimer.purge();
      }
    }
LauncherBackend类：
  static final String ENV_LAUNCHER_PORT = "_SPARK_LAUNCHER_PORT";
  static final String ENV_LAUNCHER_SECRET = "_SPARK_LAUNCHER_SECRET";

val port = sys.env.get(LauncherProtocol.ENV_LAUNCHER_PORT).map(_.toInt)val secret = sys.env.get(LauncherProtocol.ENV_LAUNCHER_SECRET)
if (port != None && secret != None) {
  val s = new Socket(InetAddress.getLoopbackAddress(), port.get)
//还是通过它建立的链接LauncherConnection
  connection = new BackendConnection(s)
//网络流的写
  connection.send(new Hello(secret.get, SPARK_VERSION))
//客户端连接
  clientThread = LauncherBackend.threadFactory.newThread(connection)
  clientThread.start()
  _isConnected = true
}

形成了客户端发送的消息，服务端达到一定时间就会关闭连接。