分布式设计与开发（三）------高一致性服务ZooKeeper

• 高可用性
• 高一致性
• 高性能

• # The number of milliseconds of each tick
• tickTime=2000
• # The number of ticks that the initial
• # synchronization phase can take
• initLimit=10
• # The number of ticks that can pass between
• # sending a request and getting an acknowledgement
• syncLimit=5
• # the directory where the snapshot is stored.
• dataDir=/home/admin/zookeeper-3.2.2/data
• # the port at which the clients will connect
• clientPort=2181
• server.1=zoo1:2888:3888
• server.2=zoo2:2888:3888

• public
  static
  void main(String[] args) {
•     try {
•         ZooKeeper zk = new ZooKeeper("10.20.147.35:2181", 30000, null);
•         String name = zk.create("/company", "alibaba".getBytes(),
•                 Ids.OPEN_ACL_UNSAFE, CreateMode.PERSISTENT_SEQUENTIAL);
•         Stat stat = new Stat();
•         System.out.println(new String(zk.getData(name, null, stat)));
•         zk.setData(name, "taobao".getBytes(), stat.getVersion(), null, null);
•         System.out.println(new String(zk.getData(name, null, stat)));
•         stat = zk.exists(name, null);
•         zk.delete(name, stat.getVersion(), null, null);
•         System.out.println(new String(zk.getData(name, null, stat)));
•     } catch (Exception e) {
•         e.printStackTrace();
•     }
• }

• 选举Leader
• 同步数据

• Leader要具有最高的zxid
• 集群中大多数的机器得到响应并follow选出的Leader

• /**
• * Removes the head of the queue and returns it, blocks until it succeeds.
• * @return The former head of the queue
• * @throws NoSuchElementException
• * @throws KeeperException
• * @throws InterruptedException
• */
• public
  byte[] take() throws KeeperException, InterruptedException {
•     TreeMap<Long,String> orderedChildren;
•     // Same as for element.  Should refactor this.
•     while(true){
•         LatchChildWatcher childWatcher = new LatchChildWatcher();
•         try{
•             orderedChildren = orderedChildren(childWatcher);
•         }catch(KeeperException.NoNodeException e){
•             zookeeper.create(dir, new
  byte[0], acl, CreateMode.PERSISTENT);
•             continue;
•         }
•         if(orderedChildren.size() == 0){
•             childWatcher.await();
•             continue;
•         }
•         for(String headNode : orderedChildren.values()){
•             String path = dir +"/"+headNode;
•             try{
•                 byte[] data = zookeeper.getData(path, false, null);
•                 zookeeper.delete(path, -1);
•                 return data;
•             }catch(KeeperException.NoNodeException e){
•                 // Another client deleted the node first.
•             }
•         }
•     }
• }
• /**
• * Inserts data into queue.
• * @param data
• * @return true if data was successfully added
• */
• public
  boolean offer(byte[] data) throws KeeperException, InterruptedException{
•     for(;;){
•         try{
•             zookeeper.create(dir+"/"+prefix, data, acl, CreateMode.PERSISTENT_SEQUENTIAL);
•             return
  true;
•         }catch(KeeperException.NoNodeException e){
•             zookeeper.create(dir, newbyte[0], acl, CreateMode.PERSISTENT);
•         }
•     }
• }

利用ZooKeeper实现分布式lock，主要是通过一个Node来代表一个Lock，当一个client去拿锁的时候，会在这个Node下创建一个自增序列的child，然后通过getChildren()方式来check创建的child是不是最靠前的，如果是则拿到锁，否则就调用exist()来check第二靠前的child，并加上watch来监视。当拿到锁的child执行完后归还锁，归还锁仅仅需要删除自己创建的child，这时watch机制会通知到所有没有拿到锁的client，这些child就会根据前面所讲的拿锁规则来竞争锁。

本文转自夜&枫博客园博客，原文链接：http://www.cnblogs.com/newstart/archive/2012/05/24/2516798.html，如需转载请自行联系原作者