Zookeeper与Curator

Zookeeper与Curator

Zookeeper作为分布式应用程序的配置服务器使用，可以将动态配置的属性和信息放到zookeeper上，应用程序运行时从zookeeper上获取运行需要的配置信息，特别是，如果应用程序添加了对zk指定节点的监听，那么当zk上的配置动态变更后，会触发应用程序响应配置信息的更改。（在看下面的内容前，具体查看Hadoop权威指南的第十四章）。

1.     Zk的常用命令

Zk的启动：

./zkServer.sh start

Zk的控制台连接

./zkCli.sh -server 127.0.0.1:2181

其它略。。。

2.     ZK的java客户端连接

Zk提供了java的客户端，maven应用如下：

       <</span>dependency>

           <</span>groupId>org.apache.zookeeper</</span>groupId>

           <</span>artifactId>zookeeper</</span>artifactId>

           <</span>version>3.4.5</</span>version>

       </</span>dependency>

可以使用zookeeper的java客户端完成对zk的操作。

下面一段代码描述了从指定zk上读取上面所有节点的例子：

public class ZKTracePrint  implements Watcher{

    private static final int SESSION_TIMEOUT = 5000;

    private ZooKeeper zk;

    private CountDownLatch latch = new CountDownLatch(1);

   

   

    public void connection(String hosts) throws IOException, InterruptedException {

       zk = new ZooKeeper(hosts, SESSION_TIMEOUT, this);//连接zk

       latch.await();//等到zk连接完成

    }

   

 

 

    public void process(WatchedEvent event) {

       if(event.getState() == KeeperState.SyncConnected)//zk连接完成

           latch.countDown();//解锁    

    }

 

    public void close() throws InterruptedException{

       zk.close();

    }

 

   

    public void listTrace(String beginPath) throws IOException{

       if(beginPath == null || "".equals(beginPath) || "/".equals(beginPath))

       {

           beginPath = "/";

       }

       listTrace(beginPath, "");

      

    }

   

    public void listTrace(String path,String deep) throws IOException{

       deep = deep + "  ";

       System.out.println(path);

       try {

           List children = zk.getChildren(path, false);

           byte[] data = zk.getData(path, false, null);

           if(children.isEmpty()){

              if(data == null)

                  return;

              System.out.println(deep+new String(data));

           }else{

              if(data != null)

                  System.out.println(deep + new String(data));

              for(String child: children){

                  if(!path.equals("/"))

                     listTrace(path+"/"+child,deep + "  ");

                  else

                     listTrace(path+child,deep + "  ");

              }

           }

       } catch (KeeperException e) {

           e.printStackTrace();

       } catch (InterruptedException e) {

           e.printStackTrace();

       }

    }

   

    public void printZKTrance(String hosts) throws IOException, InterruptedException{

       ZKTracePrint zk = new ZKTracePrint();

       zk.connection(hosts);

       zk.listTrace(null);

       zk.close();

    }

   

    public static void main(String[] args) throws IOException, InterruptedException {

       ZKTracePrint tool = new ZKTracePrint();

       tool.printZKTrance("127.0.0.1:2181");

    }

}

3.     使用Curator控制zk

Curator封装了Zk的java客户端，能够更加简单的控制zk，并具有对zk的重连机制。

Curator的maven依赖如下：

       <</span>dependency>

           <</span>groupId>com.netflix.curator</</span>groupId>

           <</span>artifactId>curator-framework</</span>artifactId>

           <</span>version>1.3.2</</span>version>

       </</span>dependency>

下面将cursor的主要操作列出来，更多详细内容请参考：

https://github.com/Netflix/curator/wiki

实例代码如下：

public class CuratorTest {

    private CuratorFramework zkTools;

   

    public void curatorDemo() throws Exception{

       zkTools = CuratorFrameworkFactory

              .builder()

              .connectString("127.0.0.1:2181")

              .namespace("zk/test")

              .retryPolicy(new RetryNTimes(5,1000))

              .connectionTimeoutMs(30000)

              .build();

       zkTools.start();

      

       zkTools.create()//创建一个路径

           .creatingParentsIfNeeded()//如果指定的节点的父节点不存在，递归创建父节点

           .withMode(CreateMode.PERSISTENT)//存储类型（临时的还是持久的）

           .withACL(ZooDefs.Ids.OPEN_ACL_UNSAFE)//访问权限

           .forPath("zk/test");//创建的路径

      

       zkTools.//对路径节点赋值

           setData().

           forPath("zk/test","hello world".getBytes(Charset.forName("utf-8")));

      

       byte[] buffer = zkTools.

                         getData().

                         usingWatcher(new ZKWatch("zk/test")).forPath("zk/test");

       System.out.println(new String(buffer));

    }

   

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

       CuratorTest test = new CuratorTest();

       test.curatorDemo();

 

    }

    public class ZKWatch implements CuratorWatcher{

       private final String path;

      

       public String getPath() {

           return path;

       }

       public ZKWatch(String path) {

           this.path = path;

       }

       @Override

       public void process(WatchedEvent event) throws Exception {

           if(event.getType() == EventType.NodeDataChanged){

              byte[] data = zkTools.getData().forPath(path);

              System.out.println(path+":"+new String(data,Charset.forName("utf-8")));

           }

       }

      

    }

}