使用Curator操作ZooKeeper

我使用的zookeeper-3.4.11，所以添加dependency的版本号是2.x.x。

	org.apache.curator
	curator-recipes
	2.12.0

基本用法，创建CuratorFramework。start方法会尝试连接zookeeper。其他的新建，读，更新，删除操作一看就能明白。

public static void main(String[] args) throws Exception {
	RetryPolicy retryPolicy = new RetryOneTime(1000);
	CuratorFramework framework = CuratorFrameworkFactory.newClient("127.0.0.1:2181", retryPolicy) ;
	framework.start();

	framework.create().forPath("/p", "hello".getBytes());
	System.out.println(new String(framework.getData().forPath("/p")));
	
	framework.setData().forPath("/p", "hello2".getBytes());
	System.out.println(new String(framework.getData().forPath("/p")));

	System.out.println(framework.checkExists().forPath("/p").getDataLength());
	
	framework.delete().forPath("/p");
}

来一个复杂点的watch，运行下面代码后，重新打开客户端 zkCli.cmd， 添加节点 /p， 更新，删除等操作，均会收到通知。

public static void main(String[] args) throws Exception {
	RetryPolicy retryPolicy = new RetryOneTime(1000);
	CuratorFramework framework = CuratorFrameworkFactory.newClient("127.0.0.1:2181", retryPolicy);
	framework.start();

	framework.getCuratorListenable().addListener(new CuratorListener() {
		public void eventReceived(CuratorFramework client, CuratorEvent event) throws Exception {
			System.out.println(event.getWatchedEvent());
			if (event.getType() == CuratorEventType.WATCHED) {
				client.checkExists().watched().forPath("/p");
			}
		}
	});
	
	framework.checkExists().watched().forPath("/p");
	System.in.read();
}

实现分布式锁：

public static void main(String[] args) throws Exception {
	RetryPolicy retryPolicy = new RetryOneTime(1000);
	CuratorFramework client = CuratorFrameworkFactory.newClient("127.0.0.1:2181", retryPolicy);
	client.start();
	
	InterProcessMutex lock = new InterProcessMutex(client, "/lock");
	if (lock.acquire(10, TimeUnit.SECONDS)) {
		System.out.println("Got lock!");
		Thread.sleep(10000);
	}
	lock.release();
}

上面代码会在/lock 下面添加临时序列节点，最小的节点会成功的获取锁。在锁释放之前通过zkCli查看会看到有下面的节点：

[zk: localhost:2181(CONNECTED) 32] ls /lock
[_c_8eeaf9ce-8ce2-425b-abfe-0481c3788486-lock-0000000002]

领导选举，在方法 takeLeadership回调时，当前结点会被选中作为leader，处理一些事情后，方法结束时退出leader。接下来会重新选举leader。

private static void leader() throws Exception {
	RetryPolicy retryPolicy = new RetryOneTime(1000);
	CuratorFramework client = CuratorFrameworkFactory.newClient("127.0.0.1:2181", retryPolicy);
	client.start();
	
	LeaderSelector selector = new LeaderSelector(client, "/leader", new LeaderSelectorListener() {
		
		public void stateChanged(CuratorFramework client, ConnectionState newState) {
		}
		
		public void takeLeadership(CuratorFramework client) throws Exception {
			System.out.println("I got the leadship!");
			// do something 
			System.out.println("I leave the leadship!");
		}
	});

	selector.autoRequeue();
	selector.start();
	
	Thread.sleep(1000);
	selector.close();
}