[置顶] Spark技术内幕：Master基于ZooKeeper的High Availability（HA）源码实现

如果Spark的部署方式选择Standalone，一个采用Master/Slaves的典型架构，那么Master是有SPOF（单点故障，Single Point of Failure）。Spark可以选用ZooKeeper来实现HA。

     ZooKeeper提供了一个Leader Election机制，利用这个机制可以保证虽然集群存在多个Master但是只有一个是Active的，其他的都是Standby，当Active的Master出现故障时，另外的一个Standby Master会被选举出来。由于集群的信息，包括Worker， Driver和Application的信息都已经持久化到文件系统，因此在切换的过程中只会影响新Job的提交，对于正在进行的Job没有任何的影响。加入ZooKeeper的集群整体架构如下图所示。

1. Master的重启策略

Master在启动时，会根据启动参数来决定不同的Master故障重启策略：

1. ZOOKEEPER实现HA
2. FILESYSTEM：实现Master无数据丢失重启，集群的运行时数据会保存到本地/网络文件系统上
   
3. 丢弃所有原来的数据重启
   

Master::preStart()可以看出这三种不同逻辑的实现。

override def preStart() {
    logInfo("Starting Spark master at " + masterUrl)
    ...
    //persistenceEngine是持久化Worker，Driver和Application信息的，这样在Master重新启动时不会影响
    //已经提交Job的运行
    persistenceEngine = RECOVERY_MODE match {
      case "ZOOKEEPER" =>
        logInfo("Persisting recovery state to ZooKeeper")
        new ZooKeeperPersistenceEngine(SerializationExtension(context.system), conf)
      case "FILESYSTEM" =>
        logInfo("Persisting recovery state to directory: " + RECOVERY_DIR)
        new FileSystemPersistenceEngine(RECOVERY_DIR, SerializationExtension(context.system))
      case _ =>
        new BlackHolePersistenceEngine()
    }
    //leaderElectionAgent负责Leader的选取。
    leaderElectionAgent = RECOVERY_MODE match {
        case "ZOOKEEPER" =>
          context.actorOf(Props(classOf[ZooKeeperLeaderElectionAgent], self, masterUrl, conf))
        case _ => // 仅仅有一个Master的集群，那么当前的Master就是Active的
          context.actorOf(Props(classOf[MonarchyLeaderAgent], self))
      }
  }

RECOVERY_MODE是一个字符串，可以从spark-env.sh中去设置。

val RECOVERY_MODE = conf.get("spark.deploy.recoveryMode", "NONE")

如果不设置spark.deploy.recoveryMode的话，那么集群的所有运行数据在Master重启是都会丢失，这个结论是从BlackHolePersistenceEngine的实现得出的。

private[spark] class BlackHolePersistenceEngine extends PersistenceEngine {
  override def addApplication(app: ApplicationInfo) {}
  override def removeApplication(app: ApplicationInfo) {}
  override def addWorker(worker: WorkerInfo) {}
  override def removeWorker(worker: WorkerInfo) {}
  override def addDriver(driver: DriverInfo) {}
  override def removeDriver(driver: DriverInfo) {}

  override def readPersistedData() = (Nil, Nil, Nil)
}

它把所有的接口实现为空。PersistenceEngine是一个trait。作为对比，可以看一下ZooKeeper的实现。

class ZooKeeperPersistenceEngine(serialization: Serialization, conf: SparkConf)
  extends PersistenceEngine
  with Logging
{
  val WORKING_DIR = conf.get("spark.deploy.zookeeper.dir", "/spark") + "/master_status"
  val zk: CuratorFramework = SparkCuratorUtil.newClient(conf)

  SparkCuratorUtil.mkdir(zk, WORKING_DIR)
  // 将app的信息序列化到文件WORKING_DIR/app_{app.id}中
  override def addApplication(app: ApplicationInfo) {
    serializeIntoFile(WORKING_DIR + "/app_" + app.id, app)
  }

  override def removeApplication(app: ApplicationInfo) {
    zk.delete().forPath(WORKING_DIR + "/app_" + app.id)
  }

Spark使用的并不是ZooKeeper的API，而是使用的org.apache.curator.framework.CuratorFramework 和 org.apache.curator.framework.recipes.leader.{LeaderLatchListener, LeaderLatch} 。Curator在ZooKeeper上做了一层很友好的封装。

2. 集群启动参数的配置

简单总结一下参数的设置，通过上述代码的分析，我们知道为了使用ZooKeeper至少应该设置一下参数（实际上，仅仅需要设置这些参数。通过设置spark-env.sh：

spark.deploy.recoveryMode=ZOOKEEPER
spark.deploy.zookeeper.url=zk_server_1:2181,zk_server_2:2181
spark.deploy.zookeeper.dir=/dir
// OR 通过一下方式设置
export SPARK_DAEMON_JAVA_OPTS="-Dspark.deploy.recoveryMode=ZOOKEEPER "
export SPARK_DAEMON_JAVA_OPTS="${SPARK_DAEMON_JAVA_OPTS} -Dspark.deploy.zookeeper.url=zk_server1:2181,zk_server_2:2181"

各个参数的意义：

参数	默认值	含义
spark.deploy.recoveryMode	NONE	恢复模式（Master重新启动的模式），有三种：1, ZooKeeper, 2， FileSystem, 3 NONE
spark.deploy.zookeeper.url		ZooKeeper的Server地址
spark.deploy.zookeeper.dir	/spark	ZooKeeper 保存集群元数据信息的文件目录，包括Worker，Driver和Application。

3. CuratorFramework简介

CuratorFramework极大的简化了ZooKeeper的使用，它提供了high-level的API，并且基于ZooKeeper添加了很多特性，包括

• 自动连接管理：连接到ZooKeeper的Client有可能会连接中断，Curator处理了这种情况，对于Client来说自动重连是透明的。
• 简洁的API：简化了原生态的ZooKeeper的方法，事件等；提供了一个简单易用的接口。
• Recipe的实现（更多介绍请点击Recipes）：
  • Leader的选择
  • 共享锁
  • 缓存和监控
  • 分布式的队列
  • 分布式的优先队列

CuratorFrameworks通过CuratorFrameworkFactory来创建线程安全的ZooKeeper的实例。

CuratorFrameworkFactory.newClient()提供了一个简单的方式来创建ZooKeeper的实例，可以传入不同的参数来对实例进行完全的控制。获取实例后，必须通过start()来启动这个实例，在结束时，需要调用close()。

/**
     * Create a new client
     *
     *
     * @param connectString list of servers to connect to
     * @param sessionTimeoutMs session timeout
     * @param connectionTimeoutMs connection timeout
     * @param retryPolicy retry policy to use
     * @return client
     */
    public static CuratorFramework newClient(String connectString, int sessionTimeoutMs, int connectionTimeoutMs, RetryPolicy retryPolicy)
    {
        return builder().
            connectString(connectString).
            sessionTimeoutMs(sessionTimeoutMs).
            connectionTimeoutMs(connectionTimeoutMs).
            retryPolicy(retryPolicy).
            build();
    }

需要关注的还有两个Recipe：org.apache.curator.framework.recipes.leader.{LeaderLatchListener, LeaderLatch}。

首先看一下LeaderlatchListener，它在LeaderLatch状态变化的时候被通知：

1. 在该节点被选为Leader的时候，接口isLeader()会被调用
2. 在节点被剥夺Leader的时候，接口notLeader()会被调用

由于通知是异步的，因此有可能在接口被调用的时候，这个状态是准确的，需要确认一下LeaderLatch的hasLeadership()是否的确是true/false。这一点在接下来Spark的实现中可以得到体现。

/**
* LeaderLatchListener can be used to be notified asynchronously about when the state of the LeaderLatch has changed.
*
* Note that just because you are in the middle of one of these method calls, it does not necessarily mean that
* hasLeadership() is the corresponding true/false value. It is possible for the state to change behind the scenes
* before these methods get called. The contract is that if that happens, you should see another call to the other
* method pretty quickly.
*/
public interface LeaderLatchListener
{
  /**
* This is called when the LeaderLatch's state goes from hasLeadership = false to hasLeadership = true.
*
* Note that it is possible that by the time this method call happens, hasLeadership has fallen back to false. If
* this occurs, you can expect {@link #notLeader()} to also be called.
*/
  public void isLeader();

  /**
* This is called when the LeaderLatch's state goes from hasLeadership = true to hasLeadership = false.
*
* Note that it is possible that by the time this method call happens, hasLeadership has become true. If
* this occurs, you can expect {@link #isLeader()} to also be called.
*/
  public void notLeader();
}

LeaderLatch负责在众多连接到ZooKeeper Cluster的竞争者中选择一个Leader。Leader的选择机制可以看ZooKeeper的具体实现，LeaderLatch这是完成了很好的封装。我们只需要要知道在初始化它的实例后，需要通过

public class LeaderLatch implements Closeable
{
    private final Logger log = LoggerFactory.getLogger(getClass());
    private final CuratorFramework client;
    private final String latchPath;
    private final String id;
    private final AtomicReference<State> state = new AtomicReference<State>(State.LATENT);
    private final AtomicBoolean hasLeadership = new AtomicBoolean(false);
    private final AtomicReference<String> ourPath = new AtomicReference<String>();
    private final ListenerContainer<LeaderLatchListener> listeners = new ListenerContainer<LeaderLatchListener>();
    private final CloseMode closeMode;
    private final AtomicReference<Future<?>> startTask = new AtomicReference<Future<?>>();
.
.
.
    /**
     * Attaches a listener to this LeaderLatch
     * <p/>
     * Attaching the same listener multiple times is a noop from the second time on.
     * <p/>
     * All methods for the listener are run using the provided Executor.  It is common to pass in a single-threaded
     * executor so that you can be certain that listener methods are called in sequence, but if you are fine with
     * them being called out of order you are welcome to use multiple threads.
     *
     * @param listener the listener to attach
     */
    public void addListener(LeaderLatchListener listener)
    {
        listeners.addListener(listener);
    }

通过addListener可以将我们实现的Listener添加到LeaderLatch。在Listener里，我们在两个接口里实现了被选为Leader或者被剥夺Leader角色时的逻辑即可。

4. ZooKeeperLeaderElectionAgent的实现

实际上因为有Curator的存在，Spark实现Master的HA就变得非常简单了，ZooKeeperLeaderElectionAgent实现了接口LeaderLatchListener，在isLeader()确认所属的Master被选为Leader后，向Master发送消息ElectedLeader，Master会将自己的状态改为ALIVE。当noLeader()被调用时，它会向Master发送消息RevokedLeadership时，Master会关闭。

private[spark] class ZooKeeperLeaderElectionAgent(val masterActor: ActorRef,
    masterUrl: String, conf: SparkConf)
  extends LeaderElectionAgent with LeaderLatchListener with Logging  {
  val WORKING_DIR = conf.get("spark.deploy.zookeeper.dir", "/spark") + "/leader_election"
  // zk是通过CuratorFrameworkFactory创建的ZooKeeper实例
  private var zk: CuratorFramework = _
  // leaderLatch：Curator负责选出Leader。
  private var leaderLatch: LeaderLatch = _
  private var status = LeadershipStatus.NOT_LEADER

  override def preStart() {

    logInfo("Starting ZooKeeper LeaderElection agent")
    zk = SparkCuratorUtil.newClient(conf)
    leaderLatch = new LeaderLatch(zk, WORKING_DIR)
    leaderLatch.addListener(this)

    leaderLatch.start()
  }

在prestart中，启动了leaderLatch来处理选举ZK中的Leader。就如在上节分析的，主要的逻辑在isLeader和noLeader中。

override def isLeader() {
  synchronized {
    // could have lost leadership by now.
    //现在leadership可能已经被剥夺了。。详情参见Curator的实现。
    if (!leaderLatch.hasLeadership) {
      return
    }

    logInfo("We have gained leadership")
    updateLeadershipStatus(true)
  }
}

override def notLeader() {
  synchronized {
    // 现在可能赋予leadership了。详情参见Curator的实现。
    if (leaderLatch.hasLeadership) {
      return
    }

    logInfo("We have lost leadership")
    updateLeadershipStatus(false)
  }
}

updateLeadershipStatus的逻辑很简单，就是向Master发送消息。

def updateLeadershipStatus(isLeader: Boolean) {
    if (isLeader && status == LeadershipStatus.NOT_LEADER) {
      status = LeadershipStatus.LEADER
      masterActor ! ElectedLeader
    } else if (!isLeader && status == LeadershipStatus.LEADER) {
      status = LeadershipStatus.NOT_LEADER
      masterActor ! RevokedLeadership
    }
  }

5. 设计理念

为了解决Standalone模式下的Master的SPOF，Spark采用了ZooKeeper提供的选举功能。Spark并没有采用ZooKeeper原生的Java API，而是采用了Curator，一个对ZooKeeper进行了封装的框架。采用了Curator后，Spark不用管理与ZooKeeper的连接，这些对于Spark来说都是透明的。Spark仅仅使用了100行代码，就实现了Master的HA。当然了，Spark是站在的巨人的肩膀上。谁又会去重复发明轮子呢？