org.apache.flume.SinkProcessor 扩展了LifecycleAware, Configurable接口的接口类，操作多个sink的抽象层（类似于proxy），用来分配给SinkRunner对象
抽象方法：
process 和Sink 的process方法类似（内部实现增加了选择Sink的功能）
setSinks 设置sinks
具体实现类：
org.apache.flume.sink.SinkProcessorFactory 设计模式的工厂模式，用于返回SinkProcessor对象（比如SinkGroup中就会调用这个类的getProcessor方法返回SinkProcessor对象）
提供getProcessor方法，根据type的设置和SinkProcessorType返回SinkProcessor对象，并使用processor.setSinks(sinks);设置Sink列表
其中org.apache.flume.conf.sink.SinkProcessorType是一个enum类，定义了processor type到类名的对应关系：
OTHER->null FAILOVER->org.apache.flume.sink.FailoverSinkProcessor DEFAULT->org.apache.flume.sink.DefaultSinkProcessor //默认 LOAD_BALANCE->org.apache.flume.sink.LoadBalancingSinkProcessor
org.apache.flume.sink.AbstractSinkProcessor实现了SinkProcessor接口
org.apache.flume.sink.DefaultSinkProcessor实现了SinkProcessor和ConfigurableComponent接口,在没有使用sink group时使用的processor，不会做额外的操作，只是简单的proxy的操作（直接process）
@Override
public Status process() throws EventDeliveryException {
return sink.process(); //直接调用Sink.process方法
}
@Override
public void setSinks(List<Sink> sinks) {
Preconditions.checkNotNull(sinks);
Preconditions.checkArgument(sinks.size() == 1, "DefaultSinkPolicy can "
+ "only handle one sink, "
+ "try using a policy that supports multiple sinks" ); //同时在setSinks方法中会检测对应的Sink是否为1个，如果不为1个会报错
sink = sinks.get(0);
}实现类org.apache.flume.sink.FailoverSinkProcessor //AbstractSinkProcessor的子类
原理：
内部定义了两个容器，分布存储存活的Sink和失败的Sink，如果失败的Sink中的元素过了cooldown时间，会调用其process方法，判断是否已经恢复正常，如果已经恢复正常就会加入到存活的Sink中，在存活的Sink中，会根据priority进行排序，并获取priority最大的那个Sink作为active sink
setSinks必须在configure之前运行，运行过程中不能添加sink
使用时，需要配置：
1）设置sinkgroups
2）设置sinkgroups的processor.type 为 failover
3）为每一个sink设置惩罚因子 processor.priority，并且设置值是唯一的（zmap中key是priority,value是sink，需要重priority查找sink），这一点比较重要
4）可以设置上线failover时间 processor.maxpenalty（默认30000s）
例子：
host1.sinkgroups = group1 host1.sinkgroups.group1.sinks = sink1 sink2 host1.sinkgroups.group1.processor.type = failover host1.sinkgroups.group1.processor.priority.sink1 = 5 host1.sinkgroups.group1.processor.priority.sink2 = 10 host1.sinkgroups.group1.processor.maxpenalty = 10000
源码分析：
首先定义了一个内部类FailedSink实现了Comparable接口，包含了sink和priority等信息，可以用来做Sink的对比排序操作
在configure方法中定义了两个容器分布用来存放 live的sink和failed的sink
private static final String PRIORITY_PREFIX = "priority.";
private static final String MAX_PENALTY_PREFIX = "maxpenalty";
private Map<String, Sink> sinks;
private Sink activeSink;
private SortedMap<Integer, Sink> liveSinks; //存放 live的sink，key是priority,value是sink
private Queue<FailedSink> failedSinks; //存放failed的sink
private int maxPenalty ;
@Override
public void configure(Context context) {
liveSinks = new TreeMap<Integer, Sink>(); //使用TreeMap存储priority到sink的对应关系， TreeMap是一个按key排序的map( 默认的排序为升序 )
failedSinks = new PriorityQueue<FailedSink>(); //使用优先级队列
Integer nextPrio = 0;
String maxPenaltyStr = context.getString( MAX_PENALTY_PREFIX); //获取设置的最大的maxpenalty 时间
if(maxPenaltyStr == null) {
maxPenalty = DEFAULT_MAX_PENALTY; //如果没有设置值，使用默认值30000
} else {
try {
maxPenalty = Integer.parseInt(maxPenaltyStr);
} catch (NumberFormatException e) {
logger.warn("{} is not a valid value for {}" ,
new Object[] { maxPenaltyStr, MAX_PENALTY_PREFIX });
maxPenalty = DEFAULT_MAX_PENALTY ; //如果设置格式错误，也使用默认值
}
}
for (Entry<String, Sink> entry : sinks.entrySet()) {
String priStr = PRIORITY_PREFIX + entry.getKey();
Integer priority;
try {
priority = Integer.parseInt(context.getString(priStr)); //从配置信息中获取每个sink的priority值
} catch (Exception e) {
priority = --nextPrio;
}
if(!liveSinks.containsKey(priority)) { // 查看liveSinks是否含有这个priority设置的项
liveSinks.put(priority, sinks.get(entry.getKey()));
} else {
logger.warn("Sink {} not added to FailverSinkProcessor as priority" +
"duplicates that of sink {}", entry.getKey(),
liveSinks.get(priority));
}
}
activeSink = liveSinks.get(liveSinks.lastKey()); // 获取最后一个Sink作为active sink(即priority最大的Sink)
}process方法：
public Status process() throws EventDeliveryException {
// Retry any failed sinks that have gone through their "cooldown " period
Long now = System.currentTimeMillis();
while(! failedSinks.isEmpty() && failedSinks.peek().getRefresh() < now) { //peek方法用于检索该队列的头部,但不会将其删除,如果此队列为空，则返回 null
FailedSink cur = failedSinks.poll(); //获取并移除此队列的头，如果此队列为空，则返回 null
Status s;
try {
s = cur.getSink().process(); //调用对应的process方法
if (s == Status.READY) { //如果sink处于READY状态
liveSinks.put(cur.getPriority(), cur.getSink()); //则插入到liveSinks 中
activeSink = liveSinks .get(liveSinks .lastKey()); //并尝试获取一次activeSink
logger.debug("Sink {} was recovered from the fail list" ,
cur.getSink().getName());
} else {
// if it‘s a backoff it needn‘t be penalized.
failedSinks.add(cur); //否则继续加入到failedSinks 中
}
return s;
} catch (Exception e) {
cur.incFails();
failedSinks.add(cur);
}
}
Status ret = null;
while( activeSink != null ) {
try {
ret = activeSink.process(); //对activeSink 调用process方法
return ret;
} catch (Exception e) {
logger.warn("Sink {} failed and has been sent to failover list" ,
activeSink.getName(), e);
activeSink = moveActiveToDeadAndGetNext(); //moveActiveToDeadAndGetNext用于从liveSinks中取出最后一个并添加到failedSinks中，同时获取新的最后一项
}
}
throw new EventDeliveryException("All sinks failed to process, " +
"nothing left to failover to");
}moveActiveToDeadAndGetNext方法用于从liveSinks中取出最后一个并添加到failedSinks中，同时获取新的最后一项
private Sink moveActiveToDeadAndGetNext() {
Integer key = liveSinks.lastKey(); //获取当前activeSink 的priority值
failedSinks.add( new FailedSink(key, activeSink , 1));
liveSinks.remove(key); //从liveSinks中删除这一项
if(liveSinks.isEmpty()) return null ;
if(liveSinks.lastKey() != null) {
return liveSinks.get(liveSinks .lastKey()); //取出新的最后一项
} else {
return null;
}
}org.apache.flume.sink.LoadBalancingSinkProcessors//AbstractSinkProcessor的子类
两种选择方式ROUND_ROBIN（默认）/RANDOM，可以自定义自己的selector，只要实现SinkSelector 接口即可。
在process方法中循环调用每一个sink(createSinkIterator返回的迭代器)，知道遇到可以正确返回的Sink并退出循环，如果所有的sink都不可用则抛出异常，默认时backoff的设置为false，这导致每一次循环都会检测所有的Sink，如果设置为true，会设置失败的Sink为backoff，一段时间后再加入可用的Sink列表中
例子：
host1.sinkgroups.group1.sinks = sink1 sink2 host1.sinkgroups.group1.processor.type = load_balance host1.sinkgroups.group1.processor.selector = <selector type> //random或者round_robin，默认是round_robin，也可以实现自己的selector （实现接口即可）SinkSelector host1.sinkgroups.group1.processor.selector.selector_property = <value>
源码分析：
定义了一个内部接口类SinkSelector，主要包含的抽象方法：
createSinkIterator（返回可用的sink的迭代器）和informSinkFailed
configure方法：
private SinkSelector selector;
....
public void configure(Context context) {
Preconditions.checkState(getSinks().size() > 1,
"The LoadBalancingSinkProcessor cannot be used for a single sink. "
+ "Please configure more than one sinks and try again." ); //sink的数量必须大于1
String selectorTypeName = context.getString( CONFIG_SELECTOR,
SELECTOR_NAME_ROUND_ROBIN); //获取selector的设置，默认是ROUND_ROBIN
Boolean shouldBackOff = context.getBoolean( CONFIG_BACKOFF, false ); //获取backoff的设置，默认是false
selector = null;
if (selectorTypeName.equalsIgnoreCase( SELECTOR_NAME_ROUND_ROBIN)) { //如果设置为ROUND_ROBIN，生成RoundRobinSinkSelector对象
selector = new RoundRobinSinkSelector(shouldBackOff);
} else if (selectorTypeName.equalsIgnoreCase(SELECTOR_NAME_RANDOM)) { //如果设置为RANDOM，生成RandomOrderSinkSelector对象
selector = new RandomOrderSinkSelector(shouldBackOff);
} else {
try {
@SuppressWarnings("unchecked" )
Class<? extends SinkSelector> klass = (Class<? extends SinkSelector>)
Class.forName(selectorTypeName); //自定义的类型的获取，自定义类型需要扩展SinkSelector类
selector = klass.newInstance();
} catch (Exception ex) {
throw new FlumeException("Unable to instantiate sink selector: "
+ selectorTypeName, ex);
}
}
selector.setSinks(getSinks());
selector.configure(
new Context(context.getSubProperties(CONFIG_SELECTOR_PREFIX)));
LOGGER.debug( "Sink selector: " + selector + " initialized" );
}process方法：
public Status process() throws EventDeliveryException {
Status status = null;
Iterator<Sink> sinkIterator = selector.createSinkIterator();//调用对应SinkSelector实现类的createSinkIterator方法，返回可用的sink的迭代器
while (sinkIterator.hasNext()) { //循环调用对应每一个sink的process方法
Sink sink = sinkIterator.next();
try {
status = sink.process();
break; //如果遇到第一个可以返回status的即退出循环
} catch (Exception ex) {
selector.informSinkFailed(sink); //如果sink失败调用对应selector的informSinkFailed方法
LOGGER.warn("Sink failed to consume event. "
+ "Attempting next sink if available." , ex);
}
}
if (status == null) { //如果所有的都出现问题，才抛出异常
throw new EventDeliveryException("All configured sinks have failed" );
}
return status;
}定义两个SinkSelector的实现类：RoundRobinSinkSelector和RandomOrderSinkSelector
以RoundRobinSinkSelector为例：
createIterator-->getIndexList //返回当前活动的对象
informSinkFailed 如果backoff设置为了true才有效（默认为false），主要是设置restoreTime等FailureState属性（在getIndexList 中会使用这个属性）
本文出自 “菜光光的博客” 博客，请务必保留此出处http://caiguangguang.blog.51cto.com/1652935/1617025
原文地址：http://caiguangguang.blog.51cto.com/1652935/1617025
