zookeeper之Watcher的基本流程

Watcher 的基本流程

ZooKeeper 的 Watcher 机制，总的来说可以分为三个过程：客户端注册 Watcher、服务器处理 Watcher 和客户端回调 Watcher客户端注册 watcher 有 3 种方式，getData、exists、getChildren；以如下代码为例来分析整个触发机制的原理。

基于 zkclient 客户端发起一个数据操作

<dependency>
     <groupId>com.101tec</groupId>
     <artifactId>zkclient</artifactId>
     <version>0.10</version>
</dependency>

public static void main(String[] args) throws KeeperException, InterruptedException, IOException {
        ZooKeeper zookeeper = new ZooKeeper("192.168.13.102:2181", 4000, new Watcher() {
            @Override
            public void process(WatchedEvent event) {
                System.out.println("event.type" + event.getType());
            }
        });
        zookeeper.create("/watch", "0".getBytes(), ZooDefs.Ids.OPEN_ACL_UNSAFE, CreateMode.PERSISTENT); // 创建节点
        zookeeper.exists("/watch", true); // 注册监听
        Thread.sleep(1000);
        zookeeper.setData("/watch", "1".getBytes(), -1); // 修改节点的值触发监听
        System.in.read();
    }

ZooKeeper API 的初始化过程

ZooKeeper zookeeper=new ZooKeeper(“192.168.11.152:2181”,4000,new Watcher(){
            public void processor(WatchedEvent event){
            System.out.println(“event.type”);
        }
    });

在创建一个 ZooKeeper 客户端对象实例时，我们通过 new Watcher()向构造方法中传入一个默认的 Watcher, 这个 Watcher 将作为整个 ZooKeeper 会话期间的默认Watcher，会一直被保存在客户端 ZKWatchManager 的 defaultWatcher 中;代码如下：

public ZooKeeper(String connectString, int sessionTimeout, Watcher watcher,boolean canBeReadOnly, HostProvider aHostProvider,ZKClientConfig clientConfig) throws IOException {
    LOG.info("Initiating client connection, connectString=" + connectString+ " sessionTimeout=" + sessionTimeout + " watcher=" + watcher);
    if (clientConfig == null) {
        clientConfig = new ZKClientConfig();
    }
    this.clientConfig = clientConfig;
    watchManager = defaultWatchManager();
    watchManager.defaultWatcher = watcher; //在这里将 watcher 设置到ZKWatchManager
    ConnectStringParser connectStringParser = new ConnectStringParser(connectString);
    hostProvider = aHostProvider;
    //初始化了 ClientCnxn，并且调用 cnxn.start()方法
    cnxn = new ClientCnxn(connectStringParser.getChrootPath(),
    hostProvider, sessionTimeout, this, watchManager,getClientCnxnSocket(), canBeReadOnly);
    cnxn.start();
}

ClientCnxn:是 Zookeeper 客户端和 Zookeeper 服务器端进行通信和事件通知处理的主要类，它内部包含两个类，

　　1. SendThread ：负责客户端和服务器端的数据通信, 也包括事件信息的传输

　　2. EventThread : 主要在客户端回调注册的 Watchers 进行通知处理

ClientCnxn 初始化

public ClientCnxn(String chrootPath, HostProvider hostProvider, int sessionTimeout, ZooKeeper zooKeeper,
             ClientWatchManager watcher, ClientCnxnSocket clientCnxnSocket,long sessionId, byte[] sessionPasswd, boolean canBeReadOnly)
    {
        this.zooKeeper = zooKeeper;
        this.watcher = watcher;
        this.sessionId = sessionId;
        this.sessionPasswd = sessionPasswd;
        this.sessionTimeout = sessionTimeout;
        this.hostProvider = hostProvider;
        this.chrootPath = chrootPath;
        connectTimeout = sessionTimeout / hostProvider.size();
        readTimeout = sessionTimeout * 2 / 3;
        readOnly = canBeReadOnly;
        sendThread = new SendThread(clientCnxnSocket); //初始化 sendThread
        eventThread = new EventThread(); //初始化 eventThread
        this.clientConfig=zooKeeper.getClientConfig();
     }
 
    public void start() { //启动两个线程
        sendThread.start();
        eventThread.start();
    }