Java版Flink（十一）时间语义和watermark-526互联

Java版Flink（十一）时间语义和watermark

一、时间语义

在这里插入图片描述
在Flink 中涉及到三个重要时间概念：EventTime、IngestionTime、ProcessingTime。

1.1、EventTime

EventTime 表示日志事件产生的时间戳，每一条数据都会记录自己生产的时间。

1.2、IngestionTime

IngestionTime 表示数据进入 Flink程序的时间

1.3、ProcessingTime

ProcessingTime 表示数据被计算处理时间，默认Flink时间属性就是ProcessingTime。

1.4、总结

一般来说，EventTime比较重要，并且实际应用也多。比如说，在游戏里，有一个关卡要求在一分钟内完成才能过关，一个用户一进入关卡就会产生一条记录并打上该记录的产生时间戳，发往服务器。但是该用户正好处于搭地跌（假设网络信号不好），假设用户已经在搭地铁过程中完成任务，并产生相对应的数据（带时间戳）。假设用户搭完地铁出站，但是此过程中维持了两三分钟，如果采用服务器的系统时间，那么程序就会判定该用户并没有通过该关卡。如果采用日志产生时间，程序会比较两条数据的时间戳之差是否在一分钟内，如果是，在通过。

二、引入 EventTime

在 Flink 的流式处理中，绝大部分的业务都会使用 eventTime，代码中引入，如下：

StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
        /**
         * 参数 TimeCharacteristic 有三种类型：
         * ProcessingTime,
         * IngestionTime,
         * EventTime;
         */
        env.setStreamTimeCharacteristic(TimeCharacteristic.EventTime);
1
2
3
4
5
6
7
8

三、引入 watermark

3.1、基本概念

日志数据产生经过网络传输，流经source，到transform，这过程中是需要时间的，大部分的数据都是按照事件时间顺序来的，但是也不排除网络等其他因素，导致数据的乱序。所谓乱序，就是指 Flink 接收到的事件的先后顺序不是严格按照事件的 Event Time 顺序排列的。比如说
理想情况下：
7，6，5，4，3，2，1（数据是从1开始的）
实际情况下：
6，3，5，7，4，2，1
那么此时出现一个问题，一旦出现乱序，如果只根据 eventTime 决定 window 的运行，我们不能明确数据是否全部到位，但又不能无限期的等下去，此时必须要有个机制来保证一个特定的时间后，必须触发 window 去进行计算了，这个特别的机制，就是 Watermark。

3.2、watermark 作用

Watermark 是一种衡量 Event Time 进展的机制。
Watermark 是用于处理乱序事件的，而正确的处理乱序事件，通常用Watermark 机制结合 window 来实现。

3.3、watermark 案例

比如说：统计滚动窗口大小为5秒（时间戳的EventTime）
第1条数据的时间戳是：101
第2条数据的时间戳是：104
第3条数据的时间戳是：105（如果不做任何处理，已经触发窗口函数计算，那么数据结果就会不正确）
第4条数据的时间戳是：102
第5条数据的时间戳是：103
第6条数据的时间戳是：107
第7条数据的时间戳是：106
第8条数据的时间戳是：110 （如果也不做任何处理，已经触发函数计算，那么数据结果还是不正确）

上面就是乱序数据不做任何处理的情况，导致数据不准确。
如何解决？
可以让窗口函数延迟一会在计算，至于多久，具体情况具体分析，还是要看数据的。
如何延迟？（假设窗口触发计算范围为 [100-105)，[105-110) 这些窗口范围不是根据第一条数据的EventTime计算的，而是根据 1979-01-01 00:00:00 开始计算，比如你要统计1个小时的窗口那么窗口大小只能是这种情况 [00:00,01:00）[01:00,02:00) )
此时可以采用watermark来延迟窗口触发计算。基于上面的数据情况可以延迟2秒触发窗口计算（后面有具体案例代码实现）。
假设第一条数据来了，数据时间戳为 101，那么此时的最大maxEvenTime就是 101，那么此时的watermark=maxEventTime - 延迟时间 2s = 99
第二条数据来了，数据时间戳为 104，那么此时的最大maxEventTime为 102，watermark = maxEventTime - 延迟时间 2s = 102 ，watermrk 表示时间戳102之前的数据全部来了，也就是[100,102)间的数据全部到了
第三条数据来了，数据时间戳为105，那么此时的最大 maxEventTime 就是105，watermark = 103。正常情况下第三条数据来了就开始执行窗口计算，但是已经设置了watermark水位线
第四条数据来了，数据时间戳为102，那么此时的最大 maxTimeEvent还是105，它会比较的，然后总是保存maxEventTime。watermark 总是等于 maxEventTime - 延迟时间 2s
第五条数据来了，maxEventTime = 105 matermark = 103
第六条数据来了，数据时间戳为 107，那么此时的最大 maxEventTime = 107，watermark = 107 - 延迟 2s = 105，此时窗口[100,105)触发窗口函数，把窗口收集的数据开始计算
以此类推，下一个窗口[105,100)触发的时间条件为 maxEventTIme = 112 watermark = 112 -2

3.4、watermark 代码案例

根据传感器id计算最近5秒内的最大温度

import com.tan.flink.bean.SensorReading;
import org.apache.flink.api.common.functions.FlatMapFunction;
import org.apache.flink.api.java.tuple.Tuple;
import org.apache.flink.shaded.guava18.com.google.common.collect.Lists;
import org.apache.flink.streaming.api.TimeCharacteristic;
import org.apache.flink.streaming.api.datastream.DataStreamSource;
import org.apache.flink.streaming.api.datastream.SingleOutputStreamOperator;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.streaming.api.functions.timestamps.BoundedOutOfOrdernessTimestampExtractor;
import org.apache.flink.streaming.api.functions.windowing.WindowFunction;
import org.apache.flink.streaming.api.windowing.time.Time;
import org.apache.flink.streaming.api.windowing.windows.TimeWindow;
import org.apache.flink.util.Collector;
import java.text.SimpleDateFormat;
import java.util.ArrayList;
import java.util.Collections;
import java.util.Comparator;
import java.util.Date;
public class EventTime_Watermark {

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

StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();

env.setParallelism(1);

// 1、设置 EventTime 语义

env.setStreamTimeCharacteristic(TimeCharacteristic.EventTime);

// 2、读取数据

DataStreamSource<String> inputDataStream = env.socketTextStream("localhost", 9999);

// 3、切割数据

SingleOutputStreamOperator<SensorReading> sensorDataStream = inputDataStream.flatMap(new CustomFlatMap());

// 4、提取并设置 eventTime

/**

* 分配时间戳和watermark

* SensorReading -> 输入类型

* Time.seconds(2L）-> 延迟时间

* 本质最终的结果是 watermark

*/

SingleOutputStreamOperator<SensorReading> watermarkDataStream = sensorDataStream.assignTimestampsAndWatermarks(new BoundedOutOfOrdernessTimestampExtractor<SensorReading>(Time.seconds(2L)) {

@Override

public long extractTimestamp(SensorReading sensorReading) {

// 提取 eventTime 需要注意的是 该 eventTime 是毫秒单位，如果 sensorReading 的时间戳是以秒单位需要 乘以 1000

return sensorReading.getTimestamp();

}

});
    <span class="token comment">// 5、分组计算</span>
    SingleOutputStreamOperator<span class="token generics function"><span class="token punctuation">&lt;</span>String<span class="token punctuation">&gt;</span></span> resultDataStream <span class="token operator">=</span> watermarkDataStream<span class="token punctuation">.</span><span class="token function">keyBy</span><span class="token punctuation">(</span><span class="token string">"id"</span><span class="token punctuation">)</span>
            <span class="token punctuation">.</span><span class="token function">timeWindow</span><span class="token punctuation">(</span>Time<span class="token punctuation">.</span><span class="token function">seconds</span><span class="token punctuation">(</span><span class="token number">5</span>L<span class="token punctuation">)</span><span class="token punctuation">)</span>
            <span class="token punctuation">.</span><span class="token function">apply</span><span class="token punctuation">(</span><span class="token keyword">new</span> <span class="token class-name">CustomWindowFunction</span><span class="token punctuation">(</span><span class="token punctuation">)</span><span class="token punctuation">)</span><span class="token punctuation">;</span>
    resultDataStream<span class="token punctuation">.</span><span class="token function">print</span><span class="token punctuation">(</span><span class="token punctuation">)</span><span class="token punctuation">;</span>
    env<span class="token punctuation">.</span><span class="token function">execute</span><span class="token punctuation">(</span><span class="token punctuation">)</span><span class="token punctuation">;</span>
<span class="token punctuation">}</span>

<span class="token keyword">public</span> <span class="token keyword">static</span> <span class="token keyword">class</span> <span class="token class-name">CustomFlatMap</span> <span class="token keyword">implements</span> <span class="token class-name">FlatMapFunction</span><span class="token generics function"><span class="token punctuation">&lt;</span>String<span class="token punctuation">,</span> SensorReading<span class="token punctuation">&gt;</span></span> <span class="token punctuation">{<!-- --></span>

    <span class="token annotation punctuation">@Override</span>
    <span class="token keyword">public</span> <span class="token keyword">void</span> <span class="token function">flatMap</span><span class="token punctuation">(</span>String input<span class="token punctuation">,</span> Collector<span class="token generics function"><span class="token punctuation">&lt;</span>SensorReading<span class="token punctuation">&gt;</span></span> collector<span class="token punctuation">)</span> <span class="token keyword">throws</span> Exception <span class="token punctuation">{<!-- --></span>
        String<span class="token punctuation">[</span><span class="token punctuation">]</span> fields <span class="token operator">=</span> input<span class="token punctuation">.</span><span class="token function">split</span><span class="token punctuation">(</span><span class="token string">","</span><span class="token punctuation">)</span><span class="token punctuation">;</span>
        String id <span class="token operator">=</span> fields<span class="token punctuation">[</span><span class="token number">0</span><span class="token punctuation">]</span><span class="token punctuation">;</span>
        <span class="token keyword">long</span> timestamp <span class="token operator">=</span> Long<span class="token punctuation">.</span><span class="token function">parseLong</span><span class="token punctuation">(</span>fields<span class="token punctuation">[</span><span class="token number">1</span><span class="token punctuation">]</span><span class="token punctuation">)</span><span class="token punctuation">;</span>
        <span class="token keyword">double</span> temperature <span class="token operator">=</span> Double<span class="token punctuation">.</span><span class="token function">parseDouble</span><span class="token punctuation">(</span>fields<span class="token punctuation">[</span><span class="token number">2</span><span class="token punctuation">]</span><span class="token punctuation">)</span><span class="token punctuation">;</span>
        collector<span class="token punctuation">.</span><span class="token function">collect</span><span class="token punctuation">(</span><span class="token keyword">new</span> <span class="token class-name">SensorReading</span><span class="token punctuation">(</span>id<span class="token punctuation">,</span> timestamp<span class="token punctuation">,</span> temperature<span class="token punctuation">)</span><span class="token punctuation">)</span><span class="token punctuation">;</span>
    <span class="token punctuation">}</span>
<span class="token punctuation">}</span>

<span class="token comment">/**
 * SensorReading -&gt; 输入数据的类型
 * String -&gt; 输出数据类型
 * Tuple -&gt; key 类型
 * TimeWindow -&gt; 窗口类型
 */</span>
<span class="token keyword">public</span> <span class="token keyword">static</span> <span class="token keyword">class</span> <span class="token class-name">CustomWindowFunction</span> <span class="token keyword">implements</span> <span class="token class-name">WindowFunction</span><span class="token generics function"><span class="token punctuation">&lt;</span>SensorReading<span class="token punctuation">,</span> String<span class="token punctuation">,</span> Tuple<span class="token punctuation">,</span> TimeWindow<span class="token punctuation">&gt;</span></span> <span class="token punctuation">{<!-- --></span>

    <span class="token annotation punctuation">@Override</span>
    <span class="token keyword">public</span> <span class="token keyword">void</span> <span class="token function">apply</span><span class="token punctuation">(</span>Tuple tuple<span class="token punctuation">,</span> TimeWindow timeWindow<span class="token punctuation">,</span> Iterable<span class="token generics function"><span class="token punctuation">&lt;</span>SensorReading<span class="token punctuation">&gt;</span></span> iterable<span class="token punctuation">,</span> Collector<span class="token generics function"><span class="token punctuation">&lt;</span>String<span class="token punctuation">&gt;</span></span> collector<span class="token punctuation">)</span> <span class="token keyword">throws</span> Exception <span class="token punctuation">{<!-- --></span>
        ArrayList<span class="token generics function"><span class="token punctuation">&lt;</span>SensorReading<span class="token punctuation">&gt;</span></span> sensorReadings <span class="token operator">=</span> Lists<span class="token punctuation">.</span><span class="token function">newArrayList</span><span class="token punctuation">(</span>iterable<span class="token punctuation">.</span><span class="token function">iterator</span><span class="token punctuation">(</span><span class="token punctuation">)</span><span class="token punctuation">)</span><span class="token punctuation">;</span>
        <span class="token comment">//  排序取出最大温度</span>
        Collections<span class="token punctuation">.</span><span class="token function">sort</span><span class="token punctuation">(</span>sensorReadings<span class="token punctuation">,</span> <span class="token keyword">new</span> <span class="token class-name">Comparator</span><span class="token generics function"><span class="token punctuation">&lt;</span>SensorReading<span class="token punctuation">&gt;</span></span><span class="token punctuation">(</span><span class="token punctuation">)</span> <span class="token punctuation">{<!-- --></span>
            <span class="token annotation punctuation">@Override</span>
            <span class="token keyword">public</span> <span class="token keyword">int</span> <span class="token function">compare</span><span class="token punctuation">(</span>SensorReading o1<span class="token punctuation">,</span> SensorReading o2<span class="token punctuation">)</span> <span class="token punctuation">{<!-- --></span>
                <span class="token keyword">return</span> o1<span class="token punctuation">.</span><span class="token function">getTemperature</span><span class="token punctuation">(</span><span class="token punctuation">)</span> <span class="token operator">&gt;=</span> o2<span class="token punctuation">.</span><span class="token function">getTemperature</span><span class="token punctuation">(</span><span class="token punctuation">)</span> <span class="token operator">?</span> <span class="token number">1</span> <span class="token operator">:</span> <span class="token operator">-</span><span class="token number">1</span><span class="token punctuation">;</span>
            <span class="token punctuation">}</span>
        <span class="token punctuation">}</span><span class="token punctuation">)</span><span class="token punctuation">;</span>
        SensorReading sensorReading <span class="token operator">=</span> sensorReadings<span class="token punctuation">.</span><span class="token function">get</span><span class="token punctuation">(</span><span class="token number">0</span><span class="token punctuation">)</span><span class="token punctuation">;</span>
        Double maxTemperature <span class="token operator">=</span> sensorReading<span class="token punctuation">.</span><span class="token function">getTemperature</span><span class="token punctuation">(</span><span class="token punctuation">)</span><span class="token punctuation">;</span>

        <span class="token comment">// 获取时间窗口范围</span>
        <span class="token keyword">long</span> start <span class="token operator">=</span> timeWindow<span class="token punctuation">.</span><span class="token function">getStart</span><span class="token punctuation">(</span><span class="token punctuation">)</span><span class="token punctuation">;</span>
        <span class="token keyword">long</span> end <span class="token operator">=</span> timeWindow<span class="token punctuation">.</span><span class="token function">getEnd</span><span class="token punctuation">(</span><span class="token punctuation">)</span><span class="token punctuation">;</span>

        SimpleDateFormat sdf <span class="token operator">=</span> <span class="token keyword">new</span> <span class="token class-name">SimpleDateFormat</span><span class="token punctuation">(</span><span class="token string">"yyyy-MM-dd HH:mm:ss"</span><span class="token punctuation">)</span><span class="token punctuation">;</span>
        Date startDate <span class="token operator">=</span> <span class="token keyword">new</span> <span class="token class-name">Date</span><span class="token punctuation">(</span>start<span class="token punctuation">)</span><span class="token punctuation">;</span>
        String startDateTime <span class="token operator">=</span> sdf<span class="token punctuation">.</span><span class="token function">format</span><span class="token punctuation">(</span>startDate<span class="token punctuation">)</span><span class="token punctuation">;</span>

        Date endDate <span class="token operator">=</span> <span class="token keyword">new</span> <span class="token class-name">Date</span><span class="token punctuation">(</span>end<span class="token punctuation">)</span><span class="token punctuation">;</span>
        String endDateTime <span class="token operator">=</span> sdf<span class="token punctuation">.</span><span class="token function">format</span><span class="token punctuation">(</span>endDate<span class="token punctuation">)</span><span class="token punctuation">;</span>

        String resultStr <span class="token operator">=</span> tuple<span class="token punctuation">.</span><span class="token function">toString</span><span class="token punctuation">(</span><span class="token punctuation">)</span> <span class="token operator">+</span> <span class="token string">" 在"</span> <span class="token operator">+</span> startDateTime <span class="token operator">+</span> <span class="token string">" - "</span> <span class="token operator">+</span> endDateTime <span class="token operator">+</span> <span class="token string">" 最大温度为 "</span> <span class="token operator">+</span> maxTemperature<span class="token punctuation">;</span>
        collector<span class="token punctuation">.</span><span class="token function">collect</span><span class="token punctuation">(</span>resultStr<span class="token punctuation">)</span><span class="token punctuation">;</span>
    <span class="token punctuation">}</span>
<span class="token punctuation">}</span>

}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101

终端输入：
sensor_1,1547718201000,20
sensor_1,1547718205000,30
sensor_1,1547718207000,50
sensor_1,1547718212000,40
sensor_1,1547718217000,30
sensor_1,1547718223000,20

控制台输出：
(sensor_1) 在2019-01-17 17:43:20 - 2019-01-17 17:43:25 最大温度为 20.0
(sensor_1) 在2019-01-17 17:43:25 - 2019-01-17 17:43:30 最大温度为 30.0
(sensor_1) 在2019-01-17 17:43:30 - 2019-01-17 17:43:35 最大温度为 40.0
(sensor_1) 在2019-01-17 17:43:35 - 2019-01-17 17:43:40 最大温度为 30.0

注意的是：如果一直没有数据来，那么当前窗口就一直不会触发窗口函数计算。

四、watermark assigner

一个问题：watermark 什么时候更新？是不是一来数据就更新呢？那如果不来数据就一直不更新吗？Flink 会每隔一段时间就会更新watermark，至于什么时候更新？有两种类型：
AssignerWithPeriodicWatermarks
AssignerWithPunctuatedWatermarks
以上两个接口都继承自 TimestampAssigner。

4.1、Assigner with periodic watermarks（周期性更新）

4.1.1、概述

周期性的生成 watermark：系统会周期性的将 watermark 插入到流中(水位线也是一种特殊的事件!)。默认周期是 200 毫秒。可以使用
ExecutionConfig.setAutoWatermarkInterval()方法进行设置。
代码如下：

StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
        // 设置 EventTime 特征
        env.setStreamTimeCharacteristic(TimeCharacteristic.EventTime);
        // 设置 Watermark 周期性更新
        env.getConfig().setAutoWatermarkInterval(500L);
1
2
3
4
5

产生 watermark 的逻辑：每隔 5 秒钟，Flink 会调用
AssignerWithPeriodicWatermarks 的 getCurrentWatermark()方法。
如果方法返回一个时间戳大于之前水位的时间戳，新的 watermark 会被插入到流中。这个检查保证了水位线是单调递增的。如果方法返回的时间戳小于等于之前水位的时间戳，则不会产生新的watermark。

4.1.2、自定义周期提取时间戳 - 案例

假设每隔500毫秒，更新watermark并设置延迟时间为2秒。自定义周期提取时间戳类，需要实现AssignerWithPeriodicWatermarks接口。

import com.tan.flink.bean.SensorReading;
import com.tan.flink.source.SourceFromCustom;
import org.apache.flink.streaming.api.TimeCharacteristic;
import org.apache.flink.streaming.api.datastream.DataStreamSource;
import org.apache.flink.streaming.api.datastream.SingleOutputStreamOperator;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.streaming.api.functions.AssignerWithPeriodicWatermarks;
import org.apache.flink.streaming.api.watermark.Watermark;
import org.apache.flink.streaming.api.windowing.time.Time;
import javax.annotation.Nullable;
public class Watermark_CustomPeriodicWatermark {

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

StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();

env.setStreamTimeCharacteristic(TimeCharacteristic.EventTime);

env.getConfig().setAutoWatermarkInterval(500L);

DataStreamSource<SensorReading> inputDataStream = env.addSource(new SourceFromCustom.CustomSource());

SingleOutputStreamOperator<SensorReading> resultDataStream = inputDataStream.keyBy("id")

.assignTimestampsAndWatermarks(new CustomPeriodicWatermark())

.timeWindowAll(Time.seconds(5L))

.maxBy("temperature");

resultDataStream.print();

env.execute();

}
<span class="token comment">/**
 * SensorReading -&gt; 输入类型
 */</span>
<span class="token keyword">public</span> <span class="token keyword">static</span> <span class="token keyword">class</span> <span class="token class-name">CustomPeriodicWatermark</span> <span class="token keyword">implements</span> <span class="token class-name">AssignerWithPeriodicWatermarks</span><span class="token generics function"><span class="token punctuation">&lt;</span>SensorReading<span class="token punctuation">&gt;</span></span> <span class="token punctuation">{<!-- --></span>

    <span class="token comment">// 延迟时间</span>
    <span class="token keyword">private</span> <span class="token keyword">long</span> bound <span class="token operator">=</span> <span class="token number">2</span> <span class="token operator">*</span> <span class="token number">1000</span><span class="token punctuation">;</span>
    <span class="token comment">// 最大事件时间戳</span>
    <span class="token keyword">private</span> <span class="token keyword">long</span> maxEventTime <span class="token operator">=</span> Long<span class="token punctuation">.</span>MIN_VALUE<span class="token punctuation">;</span>

    <span class="token comment">// 返回 获取当前 watermark</span>
    <span class="token annotation punctuation">@Nullable</span>
    <span class="token annotation punctuation">@Override</span>
    <span class="token keyword">public</span> Watermark <span class="token function">getCurrentWatermark</span><span class="token punctuation">(</span><span class="token punctuation">)</span> <span class="token punctuation">{<!-- --></span>
        <span class="token keyword">return</span> <span class="token keyword">new</span> <span class="token class-name">Watermark</span><span class="token punctuation">(</span>maxEventTime <span class="token operator">-</span> bound<span class="token punctuation">)</span><span class="token punctuation">;</span>
    <span class="token punctuation">}</span>

    <span class="token comment">// 提取日志数据事件时间戳</span>
    <span class="token annotation punctuation">@Override</span>
    <span class="token keyword">public</span> <span class="token keyword">long</span> <span class="token function">extractTimestamp</span><span class="token punctuation">(</span>SensorReading sensorReading<span class="token punctuation">,</span> <span class="token keyword">long</span> lastEventTime<span class="token punctuation">)</span> <span class="token punctuation">{<!-- --></span>
        <span class="token comment">// 判断当前数据的事件时间戳是否大于当前窗口（数据集）的最大事件时间戳</span>
        <span class="token comment">// 如果是 则更新 maxEventTime 否则不做任何处理</span>
        maxEventTime <span class="token operator">=</span> Math<span class="token punctuation">.</span><span class="token function">max</span><span class="token punctuation">(</span>sensorReading<span class="token punctuation">.</span><span class="token function">getTimestamp</span><span class="token punctuation">(</span><span class="token punctuation">)</span><span class="token punctuation">,</span> maxEventTime<span class="token punctuation">)</span><span class="token punctuation">;</span>
        <span class="token comment">// 返回当前数据的时间戳</span>
        <span class="token keyword">return</span> sensorReading<span class="token punctuation">.</span><span class="token function">getTimestamp</span><span class="token punctuation">(</span><span class="token punctuation">)</span><span class="token punctuation">;</span>
    <span class="token punctuation">}</span>
<span class="token punctuation">}</span>

}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53

当前的效果和watermark代码案例相似，注意数据源不同。

4.2、Assigner with punctuated watermarks（间断性更新）

4.2.1、概述

间断式地生成 watermark。和周期性生成的方式不同，这种方式不是固定时间的，而是可以根据需要对每条数据进行筛选和处理。

4.2.2、自定义间断性提取时间戳 - 案例

根据传感器id等于sensor_1的数据，才提取相对应的watermark，插入数据流中。需要实现AssignerWithPunctuatedWatermarks接口。

import com.tan.flink.bean.SensorReading;
import com.tan.flink.source.SourceFromCustom;
import org.apache.flink.streaming.api.TimeCharacteristic;
import org.apache.flink.streaming.api.datastream.DataStreamSource;
import org.apache.flink.streaming.api.datastream.SingleOutputStreamOperator;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.streaming.api.functions.AssignerWithPunctuatedWatermarks;
import org.apache.flink.streaming.api.watermark.Watermark;
import javax.annotation.Nullable;
public class Watermark_CustomPunctuatedWatermark {

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

StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();

env.setStreamTimeCharacteristic(TimeCharacteristic.EventTime);

// 不再设置周期行性获取watermark

// env.getConfig().setAutoWatermarkInterval(500L);

DataStreamSource<SensorReading> inputDataStream = env.addSource(new SourceFromCustom.CustomSource());

SingleOutputStreamOperator<SensorReading> resultDataStream = inputDataStream.assignTimestampsAndWatermarks(new CustomPunctuatedWatermark());

resultDataStream.print();

env.execute();
<span class="token punctuation">}</span>

<span class="token keyword">public</span> <span class="token keyword">static</span> <span class="token keyword">class</span> <span class="token class-name">CustomPunctuatedWatermark</span> <span class="token keyword">implements</span> <span class="token class-name">AssignerWithPunctuatedWatermarks</span><span class="token generics function"><span class="token punctuation">&lt;</span>SensorReading<span class="token punctuation">&gt;</span></span> <span class="token punctuation">{<!-- --></span>

    <span class="token comment">// 延迟 2s</span>
    <span class="token keyword">private</span> <span class="token keyword">long</span> bound <span class="token operator">=</span> <span class="token number">2</span> <span class="token operator">*</span> <span class="token number">1000</span>L<span class="token punctuation">;</span>

    <span class="token comment">/**
     * @param lastElement      -&gt; 上一条数据
     * @param extractTimestamp -&gt; 当前数据的时间戳 根据 extractTimestamp 方法获取
     * @return
     */</span>
    <span class="token annotation punctuation">@Nullable</span>
    <span class="token annotation punctuation">@Override</span>
    <span class="token keyword">public</span> Watermark <span class="token function">checkAndGetNextWatermark</span><span class="token punctuation">(</span>SensorReading lastElement<span class="token punctuation">,</span> <span class="token keyword">long</span> extractTimestamp<span class="token punctuation">)</span> <span class="token punctuation">{<!-- --></span>
        <span class="token comment">// 如果上一条数据的id 等于 sensor_1 则更新时间戳 否则返回 null</span>
        <span class="token keyword">if</span> <span class="token punctuation">(</span><span class="token string">"sensor_1"</span><span class="token punctuation">.</span><span class="token function">equals</span><span class="token punctuation">(</span>lastElement<span class="token punctuation">.</span><span class="token function">getId</span><span class="token punctuation">(</span><span class="token punctuation">)</span><span class="token punctuation">)</span><span class="token punctuation">)</span> <span class="token punctuation">{<!-- --></span>
            <span class="token keyword">return</span> <span class="token keyword">new</span> <span class="token class-name">Watermark</span><span class="token punctuation">(</span>extractTimestamp <span class="token operator">-</span> bound<span class="token punctuation">)</span><span class="token punctuation">;</span>
        <span class="token punctuation">}</span> <span class="token keyword">else</span> <span class="token punctuation">{<!-- --></span>
            <span class="token keyword">return</span> null<span class="token punctuation">;</span>
        <span class="token punctuation">}</span>
    <span class="token punctuation">}</span>

    <span class="token comment">/**
     * @param sensorReading            -&gt; 当前数据
     * @param previousElementTimestamp -&gt; 上一条数据的事件事件戳
     * @return
     */</span>
    <span class="token annotation punctuation">@Override</span>
    <span class="token keyword">public</span> <span class="token keyword">long</span> <span class="token function">extractTimestamp</span><span class="token punctuation">(</span>SensorReading sensorReading<span class="token punctuation">,</span> <span class="token keyword">long</span> previousElementTimestamp<span class="token punctuation">)</span> <span class="token punctuation">{<!-- --></span>
        <span class="token keyword">return</span> sensorReading<span class="token punctuation">.</span><span class="token function">getTimestamp</span><span class="token punctuation">(</span><span class="token punctuation">)</span><span class="token punctuation">;</span>
    <span class="token punctuation">}</span>
<span class="token punctuation">}</span>

}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55

4.3、总结

4.3.1、有序情况下

一种简单的特殊情况是，如果我们事先得知数据流的时间戳是单调递增的，也就是说没有乱序，那我们可以使用AscendingTimestampExtractor，这个类会直接使用数据的时间戳生成 watermark。

inputDataStream.assignTimestampsAndWatermarks(new AscendingTimestampExtractor<SensorReading>() {
            @Override
            public long extractAscendingTimestamp(SensorReading sensorReading) {
                return sensorReading.getTimestamp();
            }
        });
1
2
3
4
5
6

4.3.2、无序情况下

而对于乱序数据流，如果我们能大致估算出数据流中的事件的最大延迟时间。

inputDataStream.assignTimestampsAndWatermarks(new BoundedOutOfOrdernessTimestampExtractor<SensorReading>(Time.seconds(2L)) {
            @Override
            public long extractTimestamp(SensorReading sensorReading) {
                return sensorReading.getTimestamp();
            }
        });
1
2
3
4
5
6

原文链接：https://blog.csdn.net/qq_41311979/article/details/114984348?spm=1001.2014.3001.5502

eventtime watermark flink

watermark remover cracked hitpaw

水印watermark文件abap