Flume-ng源码解析之Channel组件

如果还没看过Flume-ng源码解析之启动流程，可以点击Flume-ng源码解析之启动流程 查看

1 接口介绍

组件的分析顺序是按照上一篇中启动顺序来分析的，首先是Channel，然后是Sink，最后是Source，在开始看组件源码之前我们先来看一下两个重要的接口，一个是LifecycleAware ，另一个是NamedComponent

1.1 LifecycleAware

@InterfaceAudience.Public
@InterfaceStability.Stable
public interface LifecycleAware {
  public void start();
  public void stop();
  public LifecycleState getLifecycleState();
}

非常简单就是三个方法，start()、stop()和getLifecycleState，这个接口是flume好多类都要实现的接口，包括Flume-ng源码解析之启动流程

所中提到PollingPropertiesFileConfigurationProvider()，只要涉及到生命周期的都会实现该接口，当然组件们也是要实现的！

1.2 NamedComponent

@InterfaceAudience.Public
@InterfaceStability.Stable
public interface NamedComponent {
  public void setName(String name);
  public String getName();
}

这个没什么好讲的，就是用来设置名字的。

2 Channel

作为Flume三大核心组件之一的Channel，我们有必要来看看它的构成：

@InterfaceAudience.Public
@InterfaceStability.Stable
public interface Channel extends LifecycleAware, NamedComponent {
  public void put(Event event) throws ChannelException;
  public Event take() throws ChannelException;
  public Transaction getTransaction();
}

那么从上面的接口中我们可以看到Channel的主要功能就是put()和take()，那么我们就来看一下它的具体实现。这里我们选择MemoryChannel作为例子，但是MemoryChannel太长了，我们就截取一小段来看看

public class MemoryChannel extends BasicChannelSemantics {
    private static Logger LOGGER = LoggerFactory.getLogger(MemoryChannel.class);
    private static final Integer defaultCapacity = Integer.valueOf(100);
    private static final Integer defaultTransCapacity = Integer.valueOf(100);
    public MemoryChannel() {
    }
    ...
}

我们又看到它继承了BasicChannelSemantics ，从名字我们可以看出它是一个基础的Channel，我们继续看看看它的实现

@InterfaceAudience.Public
@InterfaceStability.Stable
public abstract class BasicChannelSemantics extends AbstractChannel {
  private ThreadLocal<BasicTransactionSemantics> currentTransaction
      = new ThreadLocal<BasicTransactionSemantics>();
  private boolean initialized = false;
  protected void initialize() {}
  protected abstract BasicTransactionSemantics createTransaction();
  @Override
  public void put(Event event) throws ChannelException {
    BasicTransactionSemantics transaction = currentTransaction.get();
    Preconditions.checkState(transaction != null,
        "No transaction exists for this thread");
    transaction.put(event);
  }
  @Override
  public Event take() throws ChannelException {
    BasicTransactionSemantics transaction = currentTransaction.get();
    Preconditions.checkState(transaction != null,
        "No transaction exists for this thread");
    return transaction.take();
  }
  @Override
  public Transaction getTransaction() {
    if (!initialized) {
      synchronized (this) {
        if (!initialized) {
          initialize();
          initialized = true;
        }
      }
    }
    BasicTransactionSemantics transaction = currentTransaction.get();
    if (transaction == null || transaction.getState().equals(
            BasicTransactionSemantics.State.CLOSED)) {
      transaction = createTransaction();
      currentTransaction.set(transaction);
    }
    return transaction;
  }
}

找了许久，终于发现了put()和take()，但是仔细一看，它们内部调用的是BasicTransactionSemantics 的put()和take()，有点失望，继续来看看BasicTransactionSemantics

public abstract class BasicTransactionSemantics implements Transaction {
  private State state;
  private long initialThreadId;
  protected void doBegin() throws InterruptedException {}
  protected abstract void doPut(Event event) throws InterruptedException;
  protected abstract Event doTake() throws InterruptedException;
  protected abstract void doCommit() throws InterruptedException;
  protected abstract void doRollback() throws InterruptedException;
  protected void doClose() {}
  protected BasicTransactionSemantics() {
    state = State.NEW;
    initialThreadId = Thread.currentThread().getId();
  }
  protected void put(Event event) {
    Preconditions.checkState(Thread.currentThread().getId() == initialThreadId,
        "put() called from different thread than getTransaction()!");
    Preconditions.checkState(state.equals(State.OPEN),
        "put() called when transaction is %s!", state);
    Preconditions.checkArgument(event != null,
        "put() called with null event!");
    try {
      doPut(event);
    } catch (InterruptedException e) {
      Thread.currentThread().interrupt();
      throw new ChannelException(e.toString(), e);
    }
  }
  protected Event take() {
    Preconditions.checkState(Thread.currentThread().getId() == initialThreadId,
        "take() called from different thread than getTransaction()!");
    Preconditions.checkState(state.equals(State.OPEN),
        "take() called when transaction is %s!", state);
    try {
      return doTake();
    } catch (InterruptedException e) {
      Thread.currentThread().interrupt();
      return null;
    }
  }
  protected State getState() {
    return state;
  }
  ...//我们这里只是讨论put和take，所以一些暂时不涉及的方法就被我干掉，有兴趣恩典朋友可以自行阅读
  protected static enum State {
    NEW, OPEN, COMPLETED, CLOSED
  }
}

又是一个抽象类，put()和take()内部调用的还是抽象方法doPut()和doTake()，看到这里，我相信没有耐心的同学已经崩溃了，但是就差最后一步了，既然是抽象类，那么最终Channel所使用的肯定是它的一个实现类，这时候我们可以回到一开始使用的MemoryChannel，到里面找找有没有线索，一看，MemoryChannel中就藏着个内部类

private class MemoryTransaction extends BasicTransactionSemantics {
    private LinkedBlockingDeque<Event> takeList;
    private LinkedBlockingDeque<Event> putList;
    private final ChannelCounter channelCounter;
    private int putByteCounter = 0;
    private int takeByteCounter = 0;
    public MemoryTransaction(int transCapacity, ChannelCounter counter) {
      putList = new LinkedBlockingDeque<Event>(transCapacity);
      takeList = new LinkedBlockingDeque<Event>(transCapacity);
      channelCounter = counter;
    }
    @Override
    protected void doPut(Event event) throws InterruptedException {
      channelCounter.incrementEventPutAttemptCount();
      int eventByteSize = (int) Math.ceil(estimateEventSize(event) / byteCapacitySlotSize);
      if (!putList.offer(event)) {
        throw new ChannelException(
            "Put queue for MemoryTransaction of capacity " +
            putList.size() + " full, consider committing more frequently, " +
            "increasing capacity or increasing thread count");
      }
      putByteCounter += eventByteSize;
    }
    @Override
    protected Event doTake() throws InterruptedException {
      channelCounter.incrementEventTakeAttemptCount();
      if (takeList.remainingCapacity() == 0) {
        throw new ChannelException("Take list for MemoryTransaction, capacity " +
            takeList.size() + " full, consider committing more frequently, " +
            "increasing capacity, or increasing thread count");
      }
      if (!queueStored.tryAcquire(keepAlive, TimeUnit.SECONDS)) {
        return null;
      }
      Event event;
      synchronized (queueLock) {
        event = queue.poll();
      }
      Preconditions.checkNotNull(event, "Queue.poll returned NULL despite semaphore " +
          "signalling existence of entry");
      takeList.put(event);
      int eventByteSize = (int) Math.ceil(estimateEventSize(event) / byteCapacitySlotSize);
      takeByteCounter += eventByteSize;
      return event;
    }
   //...依然删除暂时不需要的方法
  }

在这个类中我们可以看到doPut()和doTake()的实现方法，也明白MemoryChannel的put()和take()最终调用的是MemoryTransaction 的doPut()和doTake()。

有朋友看到这里以为这次解析就要结束了，其实好戏还在后头，Channel中还有两个重要的类ChannelProcessor和ChannelSelector，耐心地听我慢慢道来。

3 ChannelProcessor

ChannelProcessor 的作用就是执行put操作，将数据放到channel里面。每个ChannelProcessor实例都会配备一个ChannelSelector来决定event要put到那个channl当中

public class ChannelProcessor implements Configurable {
    private static final Logger LOG = LoggerFactory.getLogger(ChannelProcessor.class);
    private final ChannelSelector selector;
    private final InterceptorChain interceptorChain;
    public ChannelProcessor(ChannelSelector selector) {
        this.selector = selector;
        this.interceptorChain = new InterceptorChain();
    }
    public void initialize() {
        this.interceptorChain.initialize();
    }
    public void close() {
        this.interceptorChain.close();
    }
    public void configure(Context context) {
        this.configureInterceptors(context);
    }
    private void configureInterceptors(Context context) {
        //配置拦截器
    }
    public ChannelSelector getSelector() {
        return this.selector;
    }
    public void processEventBatch(List<Event> events) {
        ...
        while(i$.hasNext()) {
            Event optChannel = (Event)i$.next();
            List tx = this.selector.getRequiredChannels(optChannel);
	        ...//将event放到Required队列
            t1 = this.selector.getOptionalChannels(optChannel);
            Object eventQueue;
            ...//将event放到Optional队列
        }
	    ...//event的分配操作
    }
    public void processEvent(Event event) {
        event = this.interceptorChain.intercept(event);
        if(event != null) {
            List requiredChannels = this.selector.getRequiredChannels(event);
            Iterator optionalChannels = requiredChannels.iterator();
            ...//event的分配操作
            List optionalChannels1 = this.selector.getOptionalChannels(event);
            Iterator i$1 = optionalChannels1.iterator();
            ...//event的分配操作
        }
    }
}

为了简化代码，我进行了一些删除，只保留需要讲解的部分，说白了Channel中的两个写入方法，都是需要从作为参数传入的selector中获取对应的channel来执行event的put操作。接下来我们来看看ChannelSelector

4 ChannelSelector

ChannelSelector是一个接口，我们可以通过ChannelSelectorFactory来创建它的子类，Flume提供了两个实现类MultiplexingChannelSelector和ReplicatingChannelSelector。

public interface ChannelSelector extends NamedComponent, Configurable {
    void setChannels(List<Channel> var1);
    List<Channel> getRequiredChannels(Event var1);
    List<Channel> getOptionalChannels(Event var1);
    List<Channel> getAllChannels();
}

通过ChannelSelectorFactory 的create来创建，create中调用getSelectorForType来获得一个selector，通过配置文件中的type来创建相应的子类

public class ChannelSelectorFactory {
  private static final Logger LOGGER = LoggerFactory.getLogger(
      ChannelSelectorFactory.class);
  public static ChannelSelector create(List<Channel> channels,
      Map<String, String> config) {
	  ...
  }
  public static ChannelSelector create(List<Channel> channels,
      ChannelSelectorConfiguration conf) {
    String type = ChannelSelectorType.REPLICATING.toString();
    if (conf != null) {
      type = conf.getType();
    }
    ChannelSelector selector = getSelectorForType(type);
    selector.setChannels(channels);
    Configurables.configure(selector, conf);
    return selector;
  }
  private static ChannelSelector getSelectorForType(String type) {
    if (type == null || type.trim().length() == 0) {
      return new ReplicatingChannelSelector();
    }
    String selectorClassName = type;
    ChannelSelectorType  selectorType = ChannelSelectorType.OTHER;
    try {
      selectorType = ChannelSelectorType.valueOf(type.toUpperCase(Locale.ENGLISH));
    } catch (IllegalArgumentException ex) {
      LOGGER.debug("Selector type {} is a custom type", type);
    }
    if (!selectorType.equals(ChannelSelectorType.OTHER)) {
      selectorClassName = selectorType.getChannelSelectorClassName();
    }
    ChannelSelector selector = null;
    try {
      @SuppressWarnings("unchecked")
      Class<? extends ChannelSelector> selectorClass =
          (Class<? extends ChannelSelector>) Class.forName(selectorClassName);
      selector = selectorClass.newInstance();
    } catch (Exception ex) {
      throw new FlumeException("Unable to load selector type: " + type
          + ", class: " + selectorClassName, ex);
    }
    return selector;
  }
}

对于这两种Selector简单说一下：

1）MultiplexingChannelSelector

下面是一个channel selector 配置文件

agent_foo.sources.avro-AppSrv-source1.selector.type = multiplexing
agent_foo.sources.avro-AppSrv-source1.selector.header = State
agent_foo.sources.avro-AppSrv-source1.selector.mapping.CA = mem-channel-1
agent_foo.sources.avro-AppSrv-source1.selector.mapping.AZ = file-channel-2
agent_foo.sources.avro-AppSrv-source1.selector.mapping.NY = mem-channel-1 file-channel-2
agent_foo.sources.avro-AppSrv-source1.selector.optional.CA = mem-channel-1 file-channel-2
agent_foo.sources.avro-AppSrv-source1.selector.mapping.AZ = file-channel-2
agent_foo.sources.avro-AppSrv-source1.selector.default = mem-channel-1

MultiplexingChannelSelector类中定义了三个属性，用于存储不同类型的channel

    private Map<String, List<Channel>> channelMapping;
  	private Map<String, List<Channel>> optionalChannels;
  	private List<Channel> defaultChannels;

那么具体分配原则如下：

• 如果设置了maping，那么会event肯定会给指定的channel，如果同时设置了optional，也会发送给optionalchannel
• 如果没有设置maping，设置default，那么event会发送给defaultchannel，如果还同时设置了optional，那么也会发送给optionalchannel
• 如果maping和default都没指定，如果有指定option，那么会发送给optionalchannel，但是发送给optionalchannel不会进行失败重试

2）ReplicatingChannelSelector

分配原则比较简单

• 如果是replicating的话，那么如果没有指定optional，那么全部channel都有，如果某个channel指定为option的话，那么就要从requiredChannel移除，只发送给optionalchannel

5 总结：

作为一个承上启下的组件，Channel的作用就是将source来的数据通过自己流向sink，那么ChannelProcessor就起到将event put到分配好的channel中，而分配的规则是由selector决定的，flume提供的selector有multiplexing和replicating两种。所以ChannelProcessor一般都是在Source中被调用。那么Channel的take()肯定是在Sink中调用的。