Zookeeper教程：快速开始以及结合java实现分布式Barrier和Queue

一、快速开始： 

（1）概述：Zookeeper是Hadoop的一个子项目，它是分布式系统中的协调系统，可提供的服务主要有：配置服务、名字服务、分布式同步、组服务等。 

（2）使用常见：1，统一配置：把配置放在ZooKeeper的节点中维护，当配置变更时，客户端可以收到变更的通知，并应用最新的配置。2，集群管理：集群中的节点，创建ephemeral的节点，一旦断开连接，ephemeral的节点会消失，其它的集群机器可以收到消息。3，分布式锁：多个客户端发起节点创建操作，只有一个客户端创建成功，从而获得锁。 

（3）安装和配置：通过官方下载链接zookeeper 进行下载，解压后进入conf目录，新建一个zoo.conf文件，配置内容如下：

tickTime=2000
dataDir=/Users/lsq/Documents/zookeeper/zookeeper0/data
dataLogDir=/Users/lsq/Documents/zookeeper/zookeeper0/dataLog
clientPort=4399
initLimit=5
syncLimit=2

tickTime: ZooKeeper基本时间单位（ms） 

initLimit: 指定了启动zookeeper时，zookeeper实例中的随从实例同步到领导实例的初始化连接时间限制，超出时间限制则连接失败（以tickTime为时间单位）； 

syncLimit: 指定了zookeeper正常运行时，主从节点之间同步数据的时间限制，若超过这个时间限制，那么随从实例将会被丢弃 

dataDir： zookeeper存放数据的目录； 

clientPort： 用于连接客户端的端口

接下来进入bin目录启动ZooKeeper实例以及客户端连接：

./zkServer.sh start
./zkCli.sh -server localhost:4399

接下来看看集群如何配置，其实跟单机差不多，这里我们把刚刚下载的Zookeeper复制多两份，一共是三个，配置信息如下：

tickTime=2000
dataDir=/Users/lsq/Documents/zookeeper/zookeeper0/data
dataDir=/Users/lsq/Documents/zookeeper/zookeeper0/dataLog
clientPort=4399
initLimit=5
syncLimit=2
server.1=127.0.0.1:8880:9990
server.2=127.0.0.1:8881:9991
server.3=127.0.0.1:8882:9992

三个文件夹下面的zoo.conf都是这个格式，需要修改dataDir，dataDir，clientPort， 

然后在dataDir所指向的目录下面新建一个myid文件，对应server.x，比如第一个文件夹下面的myid就填入一个1，第二个就填入一个2，以此类推。接着依次启动即可。可以采用下面的命令

echo "1" > myid

二、使用java来操作ZooKeeper实例 

一门技术最重要的就算实战了，接下来的内容将围绕这一部分来讲。 

（1）首先是Znode的创建和删除 

Znode有两种类型：短暂的和持久的。短暂的znode在创建的客户端与服务器端断开（无论是明确的断开还是故障断开）连接时，该znode都会被删除；相反，持久的znode则不会

public class CreateGroup implements Watcher {
//会话延时
private static final int SESSION_TIMEOUT = 1000;
//zk对象
private ZooKeeper zk = null;
//同步计数器
private CountDownLatch countDownLatch = new CountDownLatch(1);
//客户端连接到服务器时会触发观察者进行调用
public void process(WatchedEvent event) {
if(event.getState() == KeeperState.SyncConnected){
countDownLatch.countDown();//计数器减一
}
}
public void connect(String hosts) throws IOException, InterruptedException {
zk = new ZooKeeper(hosts, SESSION_TIMEOUT, this);
countDownLatch.await();//阻塞程序继续执行
}
//创建GROUP
public void create(String groupName) throws KeeperException, InterruptedException{
String path = "/" + groupName;
//允许任何客户端对该znode进行读写,以及znode进行持久化
String createPath = zk.create(path, null, Ids.OPEN_ACL_UNSAFE, CreateMode.PERSISTENT);
System.out.println("Created "+createPath);
}
//关闭zk
public void close() throws InterruptedException{
if(zk != null){
try {
zk.close();
} catch (InterruptedException e) {
throw e;
}finally{
zk = null;
System.gc();
}
}
}
//测试主类
public static void main(String args[]){
String host = "127.0.0.1:4399";
String groupName = "test";
CreateGroup createGroup = new CreateGroup();
try {
createGroup.connect(host);
createGroup.create(groupName);
createGroup.close();
createGroup = null;
System.gc();
} catch (IOException e) {
e.printStackTrace();
} catch (InterruptedException e) {
e.printStackTrace();
} catch (KeeperException e) {
e.printStackTrace();
}
}
}

接下来把创建和销毁分离出来作为一个独立的类,以后相关操作可以直接使用

public class ConnetctionWatcher implements Watcher {
private static final int SESSION_TIMEOUT = 5000;
protected ZooKeeper zk = null;
private CountDownLatch countDownLatch = new CountDownLatch(1);
public void process(WatchedEvent event) {
KeeperState state = event.getState();
if(state == KeeperState.SyncConnected){
countDownLatch.countDown();
}
}
public void connection(String hosts) throws IOException, InterruptedException {
zk = new ZooKeeper(hosts, SESSION_TIMEOUT, this);
countDownLatch.await();
}
public void close() throws InterruptedException {
if (null != zk) {
try {
zk.close();
} catch (InterruptedException e) {
throw e;
}finally{
zk = null;
System.gc();
}
}
}
}

接下来我们看看节点如何删除

public class DeleteGroup extends ConnetctionWatcher {
public void delete(String groupName) {
String path = "/" + groupName;
try {
List<String> children = zk.getChildren(path, false);
for(String child : children){
zk.delete(path + "/" + child, -1);
}
zk.delete(path, -1);//版本号为-1，
} catch (KeeperException e) {
e.printStackTrace();
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}

git上面还有官方给的例子的代码，放在类Executor.java，DataMonitor.java，DataMonitorListener.java。有兴趣可以看一下，试着自己实现一下。

三、利用java实现分布式Barrier 

Barrier是一种控制和协调多个任务触发次序的机制。简单来说就是用一个屏障把将要执行的任务拦住，等待所有任务都处于可运行状态才放开屏障，其实在单机上我们可以利用CyclicBarrier来实现这个机制，但是在分布式环境下，我们可以利用ZooKeeper可以派上用场，我们可以利用一个Node来作为Barrier的实体，然后要Barrier的任务通过调用exists检测是否Node存在，当需要打开Barrier时候，删除这个Node，这样ZooKeeper的watch机制会通知到各个任务可以开始执行。接下来看代码：

public class Barrier extends SyncPrimitive {
int size;
String name;
Barrier(String address, String root, int size) {
super(address);
this.root = root;
this.size = size;
//创建Barrier的Node
if (zk != null) {
try {
Stat s = zk.exists(root, false);
if (s == null) {
zk.create(root, new byte[0], Ids.OPEN_ACL_UNSAFE,CreateMode.PERSISTENT);
}
} catch (KeeperException e) {
System.out.println("Keeper exception when instantiating queue: " + e.toString());
} catch (InterruptedException e) {
System.out.println("Interrupted exception");
}
}
try {
name = new String(InetAddress.getLocalHost().getCanonicalHostName().toString());
} catch (UnknownHostException e) {
System.out.println(e.toString());
}
}
/**
* 加入Barrier等待
*/
boolean enter() throws KeeperException, InterruptedException{
zk.create(root + "/" + name, new byte[0], Ids.OPEN_ACL_UNSAFE,CreateMode.EPHEMERAL_SEQUENTIAL);
while (true) {
synchronized (mutex) {
List<String> list = zk.getChildren(root, true);
if (list.size() < size) {
mutex.wait();
} else {
return true;
}
}
}
}
/**
* 一直等待知道指定数量节点到达
*/
boolean leave() throws KeeperException, InterruptedException{
zk.delete(root + "/" + name, 0);
while (true) {
synchronized (mutex) {
List<String> list = zk.getChildren(root, true);
if (list.size() > 0) {
mutex.wait();
} else {
return true;
}
}
}
}
}

父类代码如下：

public class SyncPrimitive implements Watcher {
static ZooKeeper zk = null;
static Integer mutex;
//根节点
String root;
SyncPrimitive(String address) {
if(zk == null){
try {
System.out.println("Starting ZK:");
zk = new ZooKeeper(address, 3000, this);
mutex = new Integer(-1);
System.out.println("Finished starting ZK: " + zk);
} catch (IOException e) {
System.out.println(e.toString());
zk = null;
}
}
//else mutex = new Integer(-1);
}
synchronized public void process(WatchedEvent event) {
synchronized (mutex) {
System.out.println("Process: " + event.getType());
mutex.notify();
}
}
public static void queueTest(String args[]) {
Queue q = new Queue(args[1], "/app1");
System.out.println("Input: " + args[1]);
int i;
Integer max = new Integer(args[2]);
if (args[3].equals("p")) {
System.out.println("Producer");
for (i = 0; i < max; i++)
try{
q.produce(10 + i);
} catch (KeeperException e){
} catch (InterruptedException e){
}
} else {
System.out.println("Consumer");
for (i = 0; i < max; i++) {
try{
int r = q.consume();
System.out.println("Item: " + r);
} catch (KeeperException e){
i--;
} catch (InterruptedException e){
}
}
}
}
public static void barrierTest(String args[]) {
Barrier b = new Barrier(args[1], "/b1", new Integer(args[2]));
try{
boolean flag = b.enter();
System.out.println("Entered barrier: " + args[2]);
if(!flag) System.out.println("Error when entering the barrier");
} catch (KeeperException e){
} catch (InterruptedException e){
}
Random rand = new Random();
int r = rand.nextInt(100);
for (int i = 0; i < r; i++) {
try {
Thread.sleep(100);
} catch (InterruptedException e) {
}
}
try{
b.leave();
} catch (KeeperException e){
} catch (InterruptedException e){
}
System.out.println("Left barrier");
}
//测试用的主类
public static void main(String args[]) {
/*
args =new String[] {"qTest","localhost:4399","3","c"};
if (args[0].equals("qTest"))
queueTest(args);
else
barrierTest(args);
*/
}
}

（四）分布式队列（Queue） 

在分布式环境下，实现Queue需要高一致性来保证，那么我们可以这样来设计。把一个Node当成一个队列，然后children用来存储内容，利用ZooKeeper提供的顺序递增的模式（会自动在name后面加入一个递增的数字来插入新元素）。于是在offer时候我们可以使用create，take时候按照顺序把children第一个delete就可以了。ZooKeeper保证了各个server上数据是一致的。废话不多说了，直接看代码

/**
* 一个消费者－生产者模式的消息队列
*/
public class Queue extends SyncPrimitive {
Queue(String address, String name) {
super(address);
this.root = name;
if (zk != null) {
try {
Stat s = zk.exists(root, false);
if (s == null) {
zk.create(root, new byte[0], Ids.OPEN_ACL_UNSAFE,CreateMode.PERSISTENT);
}
} catch (KeeperException e) {
System.out.println("Keeper exception when instantiating queue: " + e.toString());
} catch (InterruptedException e) {
System.out.println("Interrupted exception");
}
}
}
/**
* 队列中插入数据
*/
boolean produce(int i) throws KeeperException, InterruptedException{
ByteBuffer b = ByteBuffer.allocate(4);
byte[] value;
b.putInt(i);
value = b.array();
zk.create(root + "/element", value, Ids.OPEN_ACL_UNSAFE,CreateMode.PERSISTENT_SEQUENTIAL);
return true;
}
/**
* 把元素从队列中移除
*/
int consume() throws KeeperException, InterruptedException{
int retvalue = -1;
Stat stat = null;
//得到现在队列中首个可用的节点
while (true) {
synchronized (mutex) {
List<String> list = zk.getChildren(root, true);
if (list.size() == 0) {
System.out.println("Going to wait");
mutex.wait();
} else {
Integer min = new Integer(list.get(0).substring(7));
for(String s : list){
Integer tempValue = new Integer(s.substring(7));
//System.out.println("Temporary value: " + tempValue);
if(tempValue < min) min = tempValue;
}
System.out.println("Temporary value: " + root + "/element" + min);
byte[] b = zk.getData(root + "/element" + min, false, stat);
zk.delete(root + "/element" + min, 0);
ByteBuffer buffer = ByteBuffer.wrap(b);
retvalue = buffer.getInt();
return retvalue;
}
}
}
}
}