Apache flink 为什么PatternStream的相同事件可以同时发送到PatternSelectFunction和PatternTimeoutFunction？

我必须在给定的时间内收集3个来自kafka的流中的3个事件，这些流具有相同的correlationId，并且如果这些事件延迟到达，我能够收集全部或部分事件

我在3数据流上使用了一个union和一个CEP模式。但是我注意到，与模式非常匹配的事件，因此在选择函数中收集的事件，一旦达到超时，也会在超时函数中发送

我不知道在我的示例中我做错了什么，或者我不理解什么，但我希望积极匹配的事件也不会超时

我得到的印象是存储了不相交的时间快照

我正在使用1.3.0 Flink版本

谢谢你的帮助

控制台输出，我们可以看到3个相关事件中的2个被选中并按时间安排：

匹配事件：
键--0b3c116e-0703-43cb-8b3e-54b0b5e93948
键---f969dd4d-47ff-445c-9182-0f95a569febb
键--2ecbb89d-1463-4669-a657-555f73b6fb1d

超时事件：

首次调用超时函数：
键---f969dd4d-47ff-445c-9182-0f95a569febb
键--0b3c116e-0703-43cb-8b3e-54b0b5e93948

第二次呼叫：
键---f969dd4d-47ff-445c-9182-0f95a569febb

11:01:44,677 INFO  com.bnpp.pe.cep.Main                                          - Matching events:
11:01:44,678 INFO  com.bnpp.pe.cep.Main                                          - SctRequestProcessStep2Event(super=SctRequestEvent(correlationId=cId---a14a4e23-56c5-4242-9c43-d465d2b84454, key=Key---0b3c116e-0703-43cb-8b3e-54b0b5e93948, debtorIban=BE42063929068055, creditorIban=BE42063929068056, amount=100.0, communication=test), succeeded=false)
11:01:44,678 INFO  com.bnpp.pe.cep.Main                                          - SctRequestProcessStep1Event(super=SctRequestEvent(correlationId=cId---a14a4e23-56c5-4242-9c43-d465d2b84454, key=Key---2ecbb89d-1463-4669-a657-555f73b6fb1d, debtorIban=BE42063929068055, creditorIban=BE42063929068056, amount=100.0, communication=test), succeeded=false)
11:01:44,678 INFO  com.bnpp.pe.cep.Main                                          - SctRequestProcessStep3Event(super=SctRequestEvent(correlationId=cId---a14a4e23-56c5-4242-9c43-d465d2b84454, key=Key---f969dd4d-47ff-445c-9182-0f95a569febb, debtorIban=BE42063929068055, creditorIban=BE42063929068056, amount=100.0, communication=test), succeeded=false)
Right(SctRequestFinalEvent(super=SctRequestEvent(correlationId=cId---a14a4e23-56c5-4242-9c43-d465d2b84454, key=Key---2196fdb0-01e8-4cc6-af4b-04bcf9dc67a2, debtorIban=null, creditorIban=null, amount=null, communication=null), state=SUCCESS))
11:01:49,635 INFO  com.bnpp.pe.cep.Main                                          - Timed out events:
11:01:49,636 INFO  com.bnpp.pe.cep.Main                                          - SctRequestProcessStep3Event(super=SctRequestEvent(correlationId=cId---a14a4e23-56c5-4242-9c43-d465d2b84454, key=Key---f969dd4d-47ff-445c-9182-0f95a569febb, debtorIban=BE42063929068055, creditorIban=BE42063929068056, amount=100.0, communication=test), succeeded=false)
11:01:49,636 INFO  com.bnpp.pe.cep.Main                                          - SctRequestProcessStep2Event(super=SctRequestEvent(correlationId=cId---a14a4e23-56c5-4242-9c43-d465d2b84454, key=Key---0b3c116e-0703-43cb-8b3e-54b0b5e93948, debtorIban=BE42063929068055, creditorIban=BE42063929068056, amount=100.0, communication=test), succeeded=false)
11:01:49,636 INFO  com.bnpp.pe.cep.Main                                          - Timed out events:
11:01:49,636 INFO  com.bnpp.pe.cep.Main                                          - SctRequestProcessStep3Event(super=SctRequestEvent(correlationId=cId---a14a4e23-56c5-4242-9c43-d465d2b84454, key=Key---f969dd4d-47ff-445c-9182-0f95a569febb, debtorIban=BE42063929068055, creditorIban=BE42063929068056, amount=100.0, communication=test), succeeded=false)
Left(SctRequestFinalEvent(super=SctRequestEvent(correlationId=cId---a14a4e23-56c5-4242-9c43-d465d2b84454, key=Key---aa437bcf-ecaa-4561-9f4e-08a902f0e248, debtorIban=null, creditorIban=null, amount=null, communication=null), state=FAILED))
Left(SctRequestFinalEvent(super=SctRequestEvent(correlationId=cId---a14a4e23-56c5-4242-9c43-d465d2b84454, key=Key---5420eb41-2723-42ac-83fd-d203d6bf2526, debtorIban=null, creditorIban=null, amount=null, communication=null), state=FAILED))

我的测试代码：

package com.bnpp.pe.cep;

import com.bnpp.pe.event.Event;
import com.bnpp.pe.event.SctRequestFinalEvent;
import com.bnpp.pe.util.EventHelper;
import lombok.extern.slf4j.Slf4j;
import org.apache.flink.api.java.functions.KeySelector;
import org.apache.flink.cep.CEP;
import org.apache.flink.cep.PatternSelectFunction;
import org.apache.flink.cep.PatternStream;
import org.apache.flink.cep.PatternTimeoutFunction;
import org.apache.flink.cep.pattern.Pattern;
import org.apache.flink.streaming.api.TimeCharacteristic;
import org.apache.flink.streaming.api.datastream.DataStream;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.streaming.api.windowing.time.Time;
import org.apache.flink.streaming.connectors.kafka.FlinkKafkaConsumer010;
import org.apache.flink.streaming.util.serialization.DeserializationSchema;

import java.io.Serializable;
import java.util.List;
import java.util.Map;
import java.util.Properties;

/**
 * Created by Laurent Bauchau on 2/08/2017.
 */
@Slf4j
public class Main implements Serializable {

    public static void main(String... args) {
        new Main();
    }

    public static final String step1Topic = "sctinst-step1";
    public static final String step2Topic = "sctinst-step2";
    public static final String step3Topic = "sctinst-step3";

    private static final String PATTERN_NAME = "the_3_correlated_events_pattern";

    private final FlinkKafkaConsumer010<Event> kafkaSource1;
    private final DeserializationSchema<Event> deserializationSchema1;

    private final FlinkKafkaConsumer010<Event> kafkaSource2;
    private final DeserializationSchema<Event> deserializationSchema2;

    private final FlinkKafkaConsumer010<Event> kafkaSource3;
    private final DeserializationSchema<Event> deserializationSchema3;

    private Main() {

        // Kafka init
        Properties kafkaProperties = new Properties();
        kafkaProperties.setProperty("bootstrap.servers", "localhost:9092");
        kafkaProperties.setProperty("zookeeper.connect", "localhost:2180");
        kafkaProperties.setProperty("group.id", "sct-validation-cgroup1");

        deserializationSchema1 = new SctRequestProcessStep1EventDeserializer();
        kafkaSource1 = new FlinkKafkaConsumer010<>(step1Topic, deserializationSchema1, kafkaProperties);

        deserializationSchema2 = new SctRequestProcessStep2EventDeserializer();
        kafkaSource2 = new FlinkKafkaConsumer010<>(step2Topic, deserializationSchema2, kafkaProperties);

        deserializationSchema3 = new SctRequestProcessStep3EventDeserializer();
        kafkaSource3 = new FlinkKafkaConsumer010<>(step3Topic, deserializationSchema3, kafkaProperties);

        try {
            StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
            env.setStreamTimeCharacteristic(TimeCharacteristic.IngestionTime);

            DataStream<Event> s1 = env.addSource(kafkaSource1);
            DataStream<Event> s2 = env.addSource(kafkaSource2);
            DataStream<Event> s3 = env.addSource(kafkaSource3);

            DataStream<Event> unionStream = s1.union(s2, s3);

            Pattern successPattern = Pattern.<Event>begin(PATTERN_NAME)
                    .times(3)
                    .within(Time.seconds(5));

            PatternStream<Event> matchingStream = CEP.pattern(
                    unionStream.keyBy(new CIDKeySelector()),
                    successPattern);

            matchingStream.select(new MyPatternTimeoutFunction(), new MyPatternSelectFunction())
                    .print()
                    .setParallelism(1);

            env.execute();

        } catch (Exception e) {
            log.error(e.getMessage(), e);
        }
    }

    private static class MyPatternTimeoutFunction implements PatternTimeoutFunction<Event, SctRequestFinalEvent> {

        @Override
        public SctRequestFinalEvent timeout(Map<String, List<Event>> pattern, long timeoutTimestamp) throws Exception {

            List<Event> events = pattern.get(PATTERN_NAME);
            log.info("Timed out events:");
            events.forEach(e -> log.info(e.toString()));

            // Resulting event creation
            SctRequestFinalEvent event = new SctRequestFinalEvent();
            EventHelper.correlate(events.get(0), event);
            EventHelper.injectKey(event);
            event.setState(SctRequestFinalEvent.State.FAILED);

            return event;
        }
    }

    private static class MyPatternSelectFunction
            implements PatternSelectFunction<Event, SctRequestFinalEvent> {

        @Override
        public SctRequestFinalEvent select(Map<String, List<Event>> pattern) throws Exception {

            List<Event> events = pattern.get(PATTERN_NAME);
            log.info("Matching events:");
            events.forEach(e -> log.info(e.toString()));

            // Resulting event creation
            SctRequestFinalEvent event = new SctRequestFinalEvent();
            EventHelper.correlate(events.get(0), event);
            EventHelper.injectKey(event);
            event.setState(SctRequestFinalEvent.State.SUCCESS);

            return event;
        }
    }

    private static class CIDKeySelector implements KeySelector<Event, String> {
        @Override
        public String getKey(Event event) throws Exception {
            return event.getCorrelationId();
        }
    }
}

package com.bnpp.pe.cep；
导入com.bnpp.pe.event.event；
导入com.bnpp.pe.event.SctRequestFinalEvent；
导入com.bnpp.pe.util.EventHelper；
导入lombok.extern.slf4j.slf4j；
导入org.apache.flink.api.java.functions.KeySelector；
导入org.apache.flink.cep.cep；
导入org.apache.flink.cep.PatternSelectFunction；
导入org.apache.flink.cep.PatternStream；
导入org.apache.flink.cep.PatternTimeoutFunction；
导入org.apache.flink.cep.pattern.pattern；
导入org.apache.flink.streaming.api.TimeCharacteristic；
导入org.apache.flink.streaming.api.datastream.datastream；
导入org.apache.flink.streaming.api.environment.StreamExecutionEnvironment；
导入org.apache.flink.streaming.api.windowing.time.time；
导入org.apache.flink.streaming.connectors.kafka.FlinkKafkaConsumer010；
导入org.apache.flink.streaming.util.serialization.DeserializationSchema；
导入java.io.Serializable；
导入java.util.List；
导入java.util.Map；
导入java.util.Properties；
/**
*Laurent Bauchau于2017年8月2日创作。
*/
@Slf4j
公共类Main实现可序列化{
公共静态void main（字符串…参数）{
新的Main（）；
}
公共静态最终字符串step1Topic=“sctinst-step1”；
公共静态最终字符串step2Topic=“sctinst-step2”；
公共静态最终字符串step3Topic=“sctinst-step3”；
私有静态最终字符串模式\u NAME=“相关事件\u模式”；
私人最终FlinkKafkaConsumer010卡夫卡索资源1；
私有最终反序列化模式反序列化模式1；
私人最终FlinkKafkaConsumer010卡夫卡索资源2；
私有最终反序列化模式反序列化模式2；
私人最终FlinkKafkaConsumer010卡夫卡索资源3；
私有最终反序列化模式反序列化模式3；
专用干管（）{
//卡夫卡因特
属性kafkaProperties=新属性（）；
kafkaProperties.setProperty（“bootstrap.servers”，“localhost:9092”）；
kafkaProperties.setProperty（“zookeeper.connect”，“localhost:2180”）；
kafkaProperties.setProperty（“group.id”、“sct-validation-cgroup1”）；
反序列化Schema1=新的SctRequestProcessStep1EventDeserializer（）；
kafkaSource1=新的FlinkKafkaConsumer010（step1Topic，反序列化方案1，kafkaProperties）；
反序列化Schema2=新的SctRequestProcessStep2EventDeserializer（）；
kafkaSource2=新的FlinkKafkaConsumer010（step2Topic，反序列化模式2，kafkaProperties）；
deserializationSchema3=新的SctRequestProcessStep3EventDeserializer（）；
kafkaSource3=新的FlinkKafkaConsumer010（step3Topic，反序列化方案3，kafkaProperties）；
试一试{
StreamExecutionEnvironment env=StreamExecutionEnvironment.getExecutionEnvironment（）；
环境设置流时间特征（时间特征、摄取时间）；
数据流s1=env.addSource（kafkaSource1）；
DataStream s2=env.addSource（kafkaSource2）；
数据流s3=env.addSource（kafkaSource3）；
datastreamunionstream=s1.union（s2，s3）；
Pattern successPattern=Pattern.begin（Pattern\u NAME）
.次（3）
.在（时间.秒（5））内；
PatternStream matchingStream=CEP.pattern(
unionStream.keyBy（新的CIDKeySelector（）），
成功模式）；
matchingStream.select（新建MyPatternTimeoutFunction（），新建MyPatternSelectFunction（））
.print（）
.1（1）；
execute（）；
}捕获（例外e）{
log.error（e.getMessage（），e）；
}
}
私有静态类MyPatternTimeoutFunction实现PatternTimeoutFunction{
@凌驾
公共SctRequestFinalEvent超时（映射模式，long-timeoutTimestamp）引发异常{
列表事件=pattern.get（pattern\u NAME）；
log.info（“超时事件：”）；
events.forEach（e->log.info（e.toString（））；
//结果事件创建
SctRequestFinalEvent=新的SctRequestFinalEvent（）；
关联（events.get（0），event）；
EventHelper.injectKey（事件）；
event.setState（SctRequestFinalEvent.State.FAILED）；
返回事件；
}
}
私有静态类MyPatternSelectFunction
实现PatternSelectFunction{
@凌驾
公共SctRequestFinalEvent选择（映射模式）引发异常{
列表事件=pattern.get（pattern\u NAME）；
log.info（“匹配事件：”）；
电动汽车
Pattern.<Event>begin(PATTERN_NAME)
    .times(3)
    .within(Time.seconds(5));

import org.apache.flink.api.java.tuple.Tuple2;
import org.apache.flink.cep.CEP;
import org.apache.flink.cep.PatternSelectFunction;
import org.apache.flink.cep.PatternStream;
import org.apache.flink.cep.pattern.Pattern;
import org.apache.flink.streaming.api.datastream.DataStream;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.streaming.api.windowing.time.Time;

import java.util.Map;

public class FlinkCEP {

    public static void main(String[] args) throws Exception {

        // set up the execution environment
        final StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();

        DataStream<String> text = env.socketTextStream("localhost", 1111)
                .flatMap(new LineTokenizer());

        text.print();

        Pattern<String, String> pattern =
                Pattern.<String>begin("start").where(txt -> txt.equals("a"))
                       .next("middle").where(txt -> txt.equals("b"))
                       .followedBy("end").where(txt -> txt.equals("c")).within(Time.seconds(1));

        PatternStream<String> patternStream = CEP.pattern(text, pattern);

        DataStream<String> alerts = patternStream.select(new PatternSelectFunction<String, String>() {
            @Override
            public String select(Map<String, String> matches) throws Exception {
                return "Found: " +
                        matches.get("start") + "->" +
                        matches.get("middle") + "->" +
                        matches.get("end");
            }
        });

        // emit result
        alerts.print();

        // execute program
        env.execute("WordCount Example");
    }
}