序
本文主要研究一下flink的Triggers
Trigger
flink-streaming-java_2.11-1.7.0-sources.jar!/org/apache/flink/streaming/api/windowing/triggers/Trigger.java
@PublicEvolvingpublic abstract class Triggerimplements Serializable { private static final long serialVersionUID = -4104633972991191369L; public abstract TriggerResult onElement(T element, long timestamp, W window, TriggerContext ctx) throws Exception; public abstract TriggerResult onProcessingTime(long time, W window, TriggerContext ctx) throws Exception; public abstract TriggerResult onEventTime(long time, W window, TriggerContext ctx) throws Exception; public boolean canMerge() { return false; } public void onMerge(W window, OnMergeContext ctx) throws Exception { throw new UnsupportedOperationException("This trigger does not support merging."); } public abstract void clear(W window, TriggerContext ctx) throws Exception; // ------------------------------------------------------------------------ public interface TriggerContext { long getCurrentProcessingTime(); MetricGroup getMetricGroup(); long getCurrentWatermark(); void registerProcessingTimeTimer(long time); void registerEventTimeTimer(long time); void deleteProcessingTimeTimer(long time); void deleteEventTimeTimer(long time); S getPartitionedState(StateDescriptorstateDescriptor); @DeprecatedValueStategetKeyValueState(String name, ClassstateType, S defaultState); @DeprecatedValueStategetKeyValueState(String name, TypeInformationstateType, S defaultState); } public interface OnMergeContext extends TriggerContext {> void mergePartitionedState(StateDescriptorstateDescriptor); }}复制代码
- Trigger接收两个泛型,一个是element类型,一个是窗口类型;它定义了onElement、onProcessingTime、onEventTime、canMerge、onMerge、clear几个方法,其中onElement、onProcessingTime、onEventTime均需要返回TriggerResult
- onElement在每个element添加到window的时候会被回调;onProcessingTime在注册的event-time timer触发时会被回调;onEventTime在注册的processing-time timer触发时会被回调
- canMerge用于标识是否支持trigger state的合并,默认返回false;onMerge在多个window合并的时候会被触发;clear用于清除TriggerContext中存储的相关state
- Trigger还定义了TriggerContext及OnMergeContext;TriggerContext定义了注册及删除EventTimeTimer、ProcessingTimeTimer方法,同时还定义了getCurrentProcessingTime、getMetricGroup、getCurrentWatermark、getPartitionedState、getKeyValueState、getKeyValueState方法
- OnMergeContext继承了TriggerContext,它多定义了mergePartitionedState方法
TriggerResult
flink-streaming-java_2.11-1.7.0-sources.jar!/org/apache/flink/streaming/api/windowing/triggers/TriggerResult.java
public enum TriggerResult { CONTINUE(false, false), FIRE_AND_PURGE(true, true), FIRE(true, false), PURGE(false, true); // ------------------------------------------------------------------------ private final boolean fire; private final boolean purge; TriggerResult(boolean fire, boolean purge) { this.purge = purge; this.fire = fire; } public boolean isFire() { return fire; } public boolean isPurge() { return purge; }}复制代码
- TriggerResult用于表示trigger在onElement、onProcessingTime、onEventTime被回调时返回的action枚举,它有fire、purge两个属性,CONTINUE、FIRE_AND_PURGE、FIRE、PURGE五个枚举
- fire表示是否要触发window的computation操作;而purge表示是否要清理window的窗口数据
- CONTINUE表示不对window做任何操作;FIRE_AND_PURGE表示要触发window的computation操作然后清理window的窗口数据;FIRE表示仅仅触发window的computation操作但不清理window的窗口数据;PURGE表示不触发window的computation操作但是要清理window的窗口数据
EventTimeTrigger
flink-streaming-java_2.11-1.7.0-sources.jar!/org/apache/flink/streaming/api/windowing/triggers/EventTimeTrigger.java
@PublicEvolvingpublic class EventTimeTrigger extends Trigger
- EventTimeTrigger继承了Trigger,element类型为Object,窗口类型为TimeWindow;SlidingEventTimeWindows、TumblingEventTimeWindows、EventTimeSessionWindows、DynamicEventTimeSessionWindows默认都使用EventTimeTrigger
- onElement在window.maxTimestamp()小于等于ctx.getCurrentWatermark()的时候,返回TriggerResult.FIRE,否则执行ctx.registerEventTimeTimer(window.maxTimestamp()),然后返回TriggerResult.CONTINUE;onEventTime在time等于window.maxTimestamp()的时候返回TriggerResult.FIRE,否则返回TriggerResult.CONTINUE;onProcessingTime则返回TriggerResult.CONTINUE
- canMerge返回true;onMerge在window.maxTimestamp()大于ctx.getCurrentWatermark()的时候会执行ctx.registerEventTimeTimer(windowMaxTimestamp);clear则执行ctx.deleteEventTimeTimer(window.maxTimestamp())
ProcessingTimeTrigger
flink-streaming-java_2.11-1.7.0-sources.jar!/org/apache/flink/streaming/api/windowing/triggers/ProcessingTimeTrigger.java
@PublicEvolvingpublic class ProcessingTimeTrigger extends Trigger
- ProcessingTimeTrigger继承了Trigger,element类型为Object,窗口类型为TimeWindow;SlidingProcessingTimeWindows、TumblingProcessingTimeWindows、ProcessingTimeSessionWindows、DynamicProcessingTimeSessionWindows默认都使用ProcessingTimeTrigger
- onElement执行ctx.registerProcessingTimeTimer(window.maxTimestamp()),然后返回TriggerResult.CONTINUE;onEventTime返回TriggerResult.CONTINUE;onProcessingTime则返回TriggerResult.FIRE
- canMerge返回true;onMerge在window.maxTimestamp()大于ctx.getCurrentWatermark()的时候会执行ctx.registerProcessingTimeTimer(windowMaxTimestamp);clear则执行ctx.deleteProcessingTimeTimer(window.maxTimestamp())
NeverTrigger
flink-streaming-java_2.11-1.7.0-sources.jar!/org/apache/flink/streaming/api/windowing/assigners/GlobalWindows.java
@Internal public static class NeverTrigger extends Trigger
- NeverTrigger的onElement、onEventTime、onProcessingTime均返回TriggerResult.CONTINUE;GlobalWindows默认使用的是NeverTrigger
CountTrigger
flink-streaming-java_2.11-1.7.0-sources.jar!/org/apache/flink/streaming/api/windowing/triggers/CountTrigger.java
@PublicEvolvingpublic class CountTriggerextends Trigger
- CountTrigger继承了Trigger,指定了element类型为Object类型;它定义了maxCount及ReducingStateDescriptor;其中ReducingStateDescriptor用于窗口计数(
它使用的是自己定义的Sum函数
),在onElement方法里头,当计数大于等于maxCount时,则会清空计数,然后返回TriggerResult.FIRE,否则返回TriggerResult.CONTINUE;onEventTime、onProcessingTime均返回TriggerResult.CONTINUE;canMerge返回true;onMerge执行的是ctx.mergePartitionedState(stateDesc);clear执行的是ctx.getPartitionedState(stateDesc).clear()
PurgingTrigger
flink-streaming-java_2.11-1.7.0-sources.jar!/org/apache/flink/streaming/api/windowing/triggers/PurgingTrigger.java
@PublicEvolvingpublic class PurgingTriggerextends Trigger { private static final long serialVersionUID = 1L; private Trigger nestedTrigger; private PurgingTrigger(Trigger nestedTrigger) { this.nestedTrigger = nestedTrigger; } @Override public TriggerResult onElement(T element, long timestamp, W window, TriggerContext ctx) throws Exception { TriggerResult triggerResult = nestedTrigger.onElement(element, timestamp, window, ctx); return triggerResult.isFire() ? TriggerResult.FIRE_AND_PURGE : triggerResult; } @Override public TriggerResult onEventTime(long time, W window, TriggerContext ctx) throws Exception { TriggerResult triggerResult = nestedTrigger.onEventTime(time, window, ctx); return triggerResult.isFire() ? TriggerResult.FIRE_AND_PURGE : triggerResult; } @Override public TriggerResult onProcessingTime(long time, W window, TriggerContext ctx) throws Exception { TriggerResult triggerResult = nestedTrigger.onProcessingTime(time, window, ctx); return triggerResult.isFire() ? TriggerResult.FIRE_AND_PURGE : triggerResult; } @Override public void clear(W window, TriggerContext ctx) throws Exception { nestedTrigger.clear(window, ctx); } @Override public boolean canMerge() { return nestedTrigger.canMerge(); } @Override public void onMerge(W window, OnMergeContext ctx) throws Exception { nestedTrigger.onMerge(window, ctx); } @Override public String toString() { return "PurgingTrigger(" + nestedTrigger.toString() + ")"; } public static PurgingTrigger of(Trigger nestedTrigger) { return new PurgingTrigger<>(nestedTrigger); } @VisibleForTesting public Trigger getNestedTrigger() { return nestedTrigger; }}复制代码
- PurgingTrigger是包装型的Trigger,它包装了nestedTrigger,其onElement、onEventTime、onProcessingTime根据nestedTrigger的返回结果,在triggerResult.isFire()为true的时候,包装返回TriggerResult.FIRE_AND_PURGE;canMerge、onMerge、clear等方法均是委托给nestedTrigger处理
小结
- Trigger接收两个泛型,一个是element类型,一个是窗口类型;它定义了onElement、onProcessingTime、onEventTime、canMerge、onMerge、clear几个方法,其中onElement、onProcessingTime、onEventTime均需要返回TriggerResult;TriggerResult用于表示trigger在onElement、onProcessingTime、onEventTime被回调时返回的action枚举,它有fire、purge两个属性(
fire表示是否要触发window的computation操作;而purge表示是否要清理window的窗口数据
),CONTINUE、FIRE_AND_PURGE、FIRE、PURGE五个枚举 - SlidingEventTimeWindows、TumblingEventTimeWindows、EventTimeSessionWindows、DynamicEventTimeSessionWindows默认都使用EventTimeTrigger;SlidingProcessingTimeWindows、TumblingProcessingTimeWindows、ProcessingTimeSessionWindows、DynamicProcessingTimeSessionWindows默认都使用ProcessingTimeTrigger;GlobalWindows默认使用的是NeverTrigger
- CountTrigger主要用于计数的窗口类型,它使用ReducingStateDescriptor来进行窗口计数,在onElement方法里头,当计数大于等于maxCount时,则会清空计数,然后返回TriggerResult.FIRE,否则返回TriggerResult.CONTINUE;PurgingTrigger是包装型的Trigger,它包装了nestedTrigger,其onElement、onEventTime、onProcessingTime根据nestedTrigger的返回结果,在triggerResult.isFire()为true的时候,包装返回TriggerResult.FIRE_AND_PURGE;canMerge、onMerge、clear等方法均是委托给nestedTrigger处理