HBase中Disruptor使用

  // 核心是一个循环缓冲区。我们的循环缓冲区是一个LMAX Disruptor。当多个线程在单个WAL竞争append和sync时，它试图最小化同步与volatile写。  // Disruptor配置为处理多个生产者和仅有一个消费者（HBase中的生产者是调用append、sync的IPC Handlers）。单一的消费者从环形缓冲区中  // 拉去append和sync。

    /**     * RingBufferTruck为事件--Event     * RingBufferTruck.EVENT_FACTORY为事件工厂--EventFactory     * preallocatedEventCount为RingBuffer大小，必须是2的N次方，默认为1024*16     * appendExecutor为事件处理的线程池--newSingleThreadExecutor     * ProducerType.MULTI为生产者类型，意思为多个生产者     * BlockingWaitStrategy为指定的等待策略     */    this.disruptor =      new Disruptor<RingBufferTruck>(RingBufferTruck.EVENT_FACTORY, preallocatedEventCount,        this.appendExecutor, ProducerType.MULTI, new BlockingWaitStrategy());    // Advance the ring buffer sequence so that it starts from 1 instead of 0,    // because SyncFuture.NOT_DONE = 0.    // 请求下一个事件序号    // 让ring buffer sequence从1开始，而不是从0开始    this.disruptor.getRingBuffer().next();        // 构造事件处理器实例    this.ringBufferEventHandler =      new RingBufferEventHandler(conf.getInt("hbase.regionserver.hlog.syncer.count", 5),        maxHandlersCount);        // 添加异常处理器    this.disruptor.handleExceptionsWith(new RingBufferExceptionHandler());    // 添加事件处理器    this.disruptor.handleEventsWith(new RingBufferEventHandler [] {this.ringBufferEventHandler});        // Presize our map of SyncFutures by handler objects.    this.syncFuturesByHandler = new ConcurrentHashMap<Thread, SyncFuture>(maxHandlersCount);    // Starting up threads in constructor is a no no; Interface should have an init call.    // 启动disruptor    this.disruptor.start();