zookeeper模拟服务器端启动注册，客户端感知服务器上下线

1.首先编写服务器端程序，原理就是上线一台服务器就会在/server这个组目录下创建一个节点server，节点的值为临时的（session消失后值也会消失）且为序列的（自动分配序列号），这样服务器上线下线，zookeeper的znode中的数据也会同步上创建和删除。

服务器端模拟代码如下：

package com.lijie.zk2;

import java.security.acl.Acl;

import org.apache.zookeeper.CreateMode;
import org.apache.zookeeper.WatchedEvent;
import org.apache.zookeeper.Watcher;
import org.apache.zookeeper.ZooDefs.Ids;
import org.apache.zookeeper.ZooKeeper;
import org.apache.zookeeper.data.ACL;

public class MyServer {

    private static final String connectString = "hadoop01:2181,hadoop02:2181,hadoop03:2181";

    private static final int sessionTimeout = 2000;

    private static ZooKeeper zk = null;

    private static String group = "/server";

    public static void main(String[] args) throws Exception {

        //注册
        regServer("192.168.80.123");
        //执行任务
        business("192.168.80.123");
    }

    /**
     * 获取zookeeper实例
     * @return
     * @throws Exception
     */
    public static ZooKeeper getZookeeper() throws Exception {
        zk = new ZooKeeper(connectString, sessionTimeout, new Watcher() {
            @Override
            public void process(WatchedEvent event) {
                // 收到watch通知后的回调函数
                System.out.println("事件类型" + event.getType() + "，路径" + event.getPath());

                //因为监听器只会监听一次，这样可以一直监听,且只监听"/"目录
                try {
                    zk.getChildren("/", true);
                } catch (Exception e) {
                    // TODO Auto-generated catch block
                    e.printStackTrace();
                }
            }
        });
        return zk;
    }

    /**
     * 注册到zookeeper
     * @param ip
     * @throws Exception
     */
    public static void regServer(String ip) throws Exception {
        zk = getZookeeper();
        String create = zk.create(group + "/server", ip.getBytes(), Ids.OPEN_ACL_UNSAFE,
            CreateMode.EPHEMERAL_SEQUENTIAL);
        System.out.println(ip + " 上线了！" + ",存储路径:" + create);
    }

    /**
     * 业务
     * @throws Exception
     */
    public static void business(String ip) throws Exception {
        System.out.println(ip + " 处理业务");
        Thread.sleep(Long.MAX_VALUE);
    }
}

2.然后编写客户端程序，原理就是执行对/server下面子节点的查看并且监听，其实就相当于ls /server watch，并且在process方法中再次调用getList方法更新列表并监听，如果/server子节点有添加或者删除，那么这个监听就会被触发。

package com.lijie.zk2;

import java.util.ArrayList;
import java.util.List;

import org.apache.zookeeper.WatchedEvent;
import org.apache.zookeeper.Watcher;
import org.apache.zookeeper.ZooKeeper;

public class MyClient {

    private static final String connectString = "hadoop01:2181,hadoop02:2181,hadoop03:2181";

    private static final int sessionTimeout = 2000;

    private static ZooKeeper zk = null;

    private static String group = "/server";

    private static volatile List<String> ipList = null;

    public static void main(String[] args) throws Exception {

        //获取列表并且监听
        getList();
        //执行业务
        business();
    }

    /**
     * 获取zookeeper实例
     * @return
     * @throws Exception
     */
    public static ZooKeeper getZookeeper() throws Exception {
        zk = new ZooKeeper(connectString, sessionTimeout, new Watcher() {
            @Override
            public void process(WatchedEvent event) {
                // 收到watch通知后的回调函数
                System.out.println("事件类型" + event.getType() + "，路径" + event.getPath());

                //重新更新列表，并注册监听
                try {
                    getList();
                } catch (Exception e) {
                    // TODO Auto-generated catch block
                    e.printStackTrace();
                }
            }
        });
        return zk;
    }

    /**
     * 获取服务器列表，并监听父节点下面的变化
     * @return
     * @throws Exception
     */
    public static void getList() throws Exception {
        zk = getZookeeper();
        //获取节点名字
        List<String> children = zk.getChildren(group, true);
        //声明装载服务ip的集合
        ArrayList<String> ips = new ArrayList<String>();

        for (String path : children) {
            //获取数据
            byte[] data = zk.getData(group + "/" + path, false, null);
            ips.add(new String(data));
        }
        ipList = ips;
        //打印服务器列表
        System.out.println("打印服务器列表" + ipList);
    }

    /**
     * 业务
     * @throws Exception
     */
    public static void business() throws Exception {
        System.out.println("客户端处理");
        Thread.sleep(Long.MAX_VALUE);
    }

}

启动两个服务器端，结果如图

第一个:

第二个:

启动客户端:

注意：上面的服务器端执行没有问题，但是客户端执行的时候其实给我报错了，但是当我debug的时候却又没问题（org.apache.zookeeper.KeeperException$ConnectionLossException: KeeperErrorCode = ConnectionLoss for /server/server0000000002
），如下：

网上查找资料说是因为要先等zookeeper连接之后再执行后面的，可以使用CountDownLatch去控制，但是我这里报这个错之前会有这些ip显示出来，应该不是这个问题，而且使用CountDownLatch去控制了依旧没有效果，这个错误先放这，下次来解决。