使用Zookeeper实现分布式锁

---

实现原理

Zookeeper的一个典型应用场景就是分布式锁，锁的实现是利用Zookeeper创建的临时时序节点(创建的时候CreateMode为 EPHEMERAL_SEQUENTIAL )和节点变动的监听器实现的。时序节点保证了节点的创建在分布式系统情况下还是有先后顺序的，监听器使得客户端能感受到节点的变动情况。

---

具体步骤

1、创建一个永久性节点，作锁的根目录

2、当要获取一个锁时，在锁目录下创建一个临时有序列的节点

3、检查锁目录的子节点是否有序列比它小，若有则监听比它小的上一个节点，当前锁处于等待状态

4、当等待时间超过Zookeeper session的连接时间（sessionTimeout）时，当前session过期，Zookeeper自动删除此session创建的临时节点，等待状态结束，获取锁失败

5、当监听器触发时，等待状态结束，获得锁

---

流程图

---

详细代码

ZookeeperClient工具类

为了方便取得Zookeeper实例，先实行一个获取Zookeeper工具类 ZookeeperClient：

package org.massive.lock;

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

import java.io.IOException;
import java.util.concurrent.CountDownLatch;

/**
 * Created by Massive on 2016/12/18.
 */
public class ZookeeperClient {

    private static String connectionString = "localhost:2181";
    private static int sessionTimeout = 10000;


    public static ZooKeeper getInstance() throws IOException, InterruptedException {
        //--------------------------------------------------------------
        // 为避免连接还未完成就执行zookeeper的get/create/exists操作引起的（KeeperErrorCode = ConnectionLoss)
        // 这里等Zookeeper的连接完成才返回实例
        //--------------------------------------------------------------
        final CountDownLatch connectedSignal = new CountDownLatch(1);
        ZooKeeper zk = new ZooKeeper(connectionString, sessionTimeout, new Watcher() {
             @Override
             public void process(WatchedEvent event) {
                 if  (event.getState()  ==  Event.KeeperState.SyncConnected) {
                     connectedSignal.countDown();
                 }
             }
         });
        connectedSignal.await(sessionTimeout,TimeUnit.MILLISECONDS);
        return zk;
    }

    public static String getConnectionString() {
        return connectionString;
    }

    public static void setConnectionString(String connectionString) {
        ZookeeperClient.connectionString = connectionString;
    }

    public static int getSessionTimeout() {
        return sessionTimeout;
    }

    public static void setSessionTimeout(int sessionTimeout) {
        ZookeeperClient.sessionTimeout = sessionTimeout;
    }

}

DistributedLock

package org.massive.lock;

import org.apache.commons.lang3.RandomUtils;
import org.apache.zookeeper.*;
import org.apache.zookeeper.data.Stat;

import java.io.IOException;
import java.util.List;
import java.util.SortedSet;
import java.util.TreeSet;
import java.util.concurrent.CountDownLatch;
import java.util.concurrent.TimeUnit;

/**
 * Created by massive on 2016-12-15.
 */
public class DistributedLock {

    private String lockId;

    private static final String LOCK_ROOT = "/LOCKS";

    //--------------------------------------------------------------
    // data为存储的节点数据内容
    // 由于锁机制用的是序列功能的特性，data的值不重要，只要利于网络传输即可
    //--------------------------------------------------------------
    private final static byte[] data = {0x12, 0x34};

    private final CountDownLatch latch = new CountDownLatch(1);

    private ZooKeeper zk;

    private int sessionTimeout;

    public DistributedLock(ZooKeeper zk,int sessionTimeout) {
        this.zk = zk;
        this.sessionTimeout = sessionTimeout;
    }

    public DistributedLock() throws IOException, KeeperException, InterruptedException {
        this.zk = ZookeeperClient.getInstance();
        this.sessionTimeout = ZookeeperClient.getSessionTimeout();
    }

    class LockWatcher implements Watcher {
        @Override
        public void process(WatchedEvent event) {
            //--------------------------------------------------------------
            // 监控节点变化(本程序为序列的上一节点)
            // 若为节点删除，证明序列的上一节点已删除，此时释放阀门让当前的lock获得锁
            //--------------------------------------------------------------
            if (event.getType() == Event.EventType.NodeDeleted)
                latch.countDown();
        }
    }

    /**
     * @return
     * @throws KeeperException
     * @throws InterruptedException
     */
    public synchronized boolean lock() {

        //--------------------------------------------------------------
        // 保证锁根节点存在，若不存在则创建它
        //--------------------------------------------------------------
        createLockRootIfNotExists();

        try {

            lockId = zk.create(LOCK_ROOT + "/", data,
                    ZooDefs.Ids.OPEN_ACL_UNSAFE,
                    CreateMode.EPHEMERAL_SEQUENTIAL);

            System.out.println("thread " + Thread.currentThread().getName() +
                    " create the lock node: " + lockId + ", trying to get lock now");

            //--------------------------------------------------------------
            // 获得锁根节点下的各锁子节点，并排序
            //--------------------------------------------------------------
            List<String> nodes = zk.getChildren(LOCK_ROOT, true);
            SortedSet<String> sortedNode = new TreeSet<String>();

            for (String node : nodes) {
                sortedNode.add(LOCK_ROOT + "/" + node);
            }

            String first = sortedNode.first();
            SortedSet<String> lessThanMe = sortedNode.headSet(lockId);
            //--------------------------------------------------------------
            // 检查是否有比当前锁节点lockId更小的节点，若有则监控当前节点的前一节点
            //--------------------------------------------------------------
            if (lockId.equals(first)) {
                System.out.println("thread " + Thread.currentThread().getName() +
                        " has get the lock, lockId is " + lockId);
                return true;
            } else if (!lessThanMe.isEmpty()) {
                String prevLockId = lessThanMe.last();
                zk.exists(prevLockId, new LockWatcher());
                //--------------------------------------------------------------
                // 阀门等待sessionTimeout的时间
                // 当等待sessionTimeout的时间过后，上一个lock的Zookeeper连接会过期，删除所有临时节点，触发监听器
                //--------------------------------------------------------------
                latch.await(sessionTimeout, TimeUnit.MILLISECONDS);
                System.out.println("thread " + Thread.currentThread().getName() +
                        " has get the lock, lockId is " + lockId);
            }


        } catch (KeeperException e) {
            e.printStackTrace();
        } catch (InterruptedException e) {
            e.printStackTrace();
        }

        return true;
    }


    public synchronized boolean unlock() {
        //--------------------------------------------------------------
        // 删除lockId节点以释放锁
        //--------------------------------------------------------------
        try {
            System.out.println("thread " + Thread.currentThread().getName() +
                    " unlock the lock: " + lockId + ", the node: " + lockId + " had been deleted");
            zk.delete(lockId, -1);
            return true;
        } catch (InterruptedException e) {
            e.printStackTrace();
        } catch (KeeperException e) {
            e.printStackTrace();
        } finally {
            try {
                zk.close();
            } catch (InterruptedException e) {
                e.printStackTrace();
            }
        }
        return false;
    }

    /**
     * 保证锁根节点存在，若不存在则创建它
     */
    public void createLockRootIfNotExists() {
        try {
            Stat stat = zk.exists(LOCK_ROOT, false);
            if (stat == null) {
                zk.create(LOCK_ROOT, data, ZooDefs.Ids.OPEN_ACL_UNSAFE, CreateMode.PERSISTENT);
            }
        } catch (KeeperException e) {
            e.printStackTrace();
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
    }

    public static void main(String[] args) throws IOException, KeeperException, InterruptedException {
        final CountDownLatch latch = new CountDownLatch(10);
        for (int i = 0; i < 10; i++) {
            new Thread(new Runnable() {
                public void run() {
                    DistributedLock lock = null;
                    try {
                        lock = new DistributedLock();
                        latch.countDown();
                        latch.await();
                        lock.lock();
                        Thread.sleep(RandomUtils.nextInt(200, 500));
                    }  catch (InterruptedException e) {
                        e.printStackTrace();
                    } catch (KeeperException e) {
                        e.printStackTrace();
                    } catch (IOException e) {
                        e.printStackTrace();
                    } finally {
                        if (lock != null) {
                            lock.unlock();
                        }
                    }
                }
            }).start();
        }
    }

}

---

测试

运行main方法，本机器的某次输出结果

thread Thread-8 create the lock node: /LOCKS/0000000090, trying to get lock now
thread Thread-9 create the lock node: /LOCKS/0000000092, trying to get lock now
thread Thread-3 create the lock node: /LOCKS/0000000093, trying to get lock now
thread Thread-2 create the lock node: /LOCKS/0000000091, trying to get lock now
thread Thread-5 create the lock node: /LOCKS/0000000094, trying to get lock now
thread Thread-7 create the lock node: /LOCKS/0000000095, trying to get lock now
thread Thread-1 create the lock node: /LOCKS/0000000098, trying to get lock now
thread Thread-6 create the lock node: /LOCKS/0000000096, trying to get lock now
thread Thread-4 create the lock node: /LOCKS/0000000097, trying to get lock now
thread Thread-0 create the lock node: /LOCKS/0000000099, trying to get lock now
thread Thread-8 has get the lock, lockId is /LOCKS/0000000090
thread Thread-8 unlock the lock: /LOCKS/0000000090, the node: /LOCKS/0000000090 had been deleted
thread Thread-2 has get the lock, lockId is /LOCKS/0000000091
thread Thread-2 unlock the lock: /LOCKS/0000000091, the node: /LOCKS/0000000091 had been deleted
thread Thread-9 has get the lock, lockId is /LOCKS/0000000092
thread Thread-9 unlock the lock: /LOCKS/0000000092, the node: /LOCKS/0000000092 had been deleted
thread Thread-3 has get the lock, lockId is /LOCKS/0000000093
thread Thread-3 unlock the lock: /LOCKS/0000000093, the node: /LOCKS/0000000093 had been deleted
thread Thread-5 has get the lock, lockId is /LOCKS/0000000094
thread Thread-5 unlock the lock: /LOCKS/0000000094, the node: /LOCKS/0000000094 had been deleted
thread Thread-7 has get the lock, lockId is /LOCKS/0000000095
thread Thread-7 unlock the lock: /LOCKS/0000000095, the node: /LOCKS/0000000095 had been deleted
thread Thread-6 has get the lock, lockId is /LOCKS/0000000096
thread Thread-6 unlock the lock: /LOCKS/0000000096, the node: /LOCKS/0000000096 had been deleted
thread Thread-4 has get the lock, lockId is /LOCKS/0000000097
thread Thread-4 unlock the lock: /LOCKS/0000000097, the node: /LOCKS/0000000097 had been deleted
thread Thread-1 has get the lock, lockId is /LOCKS/0000000098
thread Thread-1 unlock the lock: /LOCKS/0000000098, the node: /LOCKS/0000000098 had been deleted
thread Thread-0 has get the lock, lockId is /LOCKS/0000000099
thread Thread-0 unlock the lock: /LOCKS/0000000099, the node: /LOCKS/0000000099 had been deleted

注：测试的结果证明了在同一JVM下，锁是正确的。由于本文锁的实现使用的是Zookeeper的监听机制，理论上在分布式系统下获取锁的顺序也是序列的顺序。但真正验证结果的正确性须在分布式系统下测试需同时运行几个JVM实例，在同一时刻去获取锁，再看输出结果。

---

参考文章：  

http://zookeeper.apache.org/doc/r3.4.9/recipes.html

http://www.ibm.com/developerworks/cn/opensource/os-cn-zookeeper/

本文为原创，转载请注明出处http://blog.csdn.net/massivestars/article/details/53771532