任务根据watermark的时间戳更新其内部事件时钟。
任务的时间服务会将所有过期的计时器标识出来,它们的时间小于当前的事件时间。对于每个过期的计时器,任务调用一个回调函数,该函数可以执行计算并发送结果。
任务会发出一个带有更新后的事件时间的watermark。
package org.bigdatatechcir.learn_flink.part5_flink_watermark;
import org.apache.flink.api.common.eventtime.TimestampAssigner;
import org.apache.flink.api.common.eventtime.TimestampAssignerSupplier;
import org.apache.flink.api.common.eventtime.Watermark;
import org.apache.flink.api.common.eventtime.WatermarkGenerator;
import org.apache.flink.api.common.eventtime.WatermarkGeneratorSupplier;
import org.apache.flink.api.common.eventtime.WatermarkOutput;
import org.apache.flink.api.common.eventtime.WatermarkStrategy;
import org.apache.flink.api.common.functions.MapFunction;
import org.apache.flink.api.common.typeinfo.Types;
import org.apache.flink.api.java.tuple.Tuple2;
import org.apache.flink.api.java.tuple.Tuple3;
import org.apache.flink.configuration.Configuration;
import org.apache.flink.configuration.RestOptions;
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.functions.source.RichParallelSourceFunction;
import org.apache.flink.streaming.api.functions.windowing.ProcessWindowFunction;
import org.apache.flink.streaming.api.windowing.assigners.TumblingEventTimeWindows;
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.time.Instant;
import java.time.ZoneId;
import java.time.ZonedDateTime;
import java.time.format.DateTimeFormatter;
import java.util.Random;
public class PunctuatedWatermarkDemo {
public static void main(String[] args) throws Exception {
Configuration conf = new Configuration();
conf.setString(RestOptions.BIND_PORT, "8081");
final StreamExecutionEnvironment env = StreamExecutionEnvironment.createLocalEnvironmentWithWebUI(conf);
env.setParallelism(4);
env.setStreamTimeCharacteristic(TimeCharacteristic.EventTime);
DataStream<String> text = env.addSource(new RichParallelSourceFunction<String>() {
private volatile boolean running = true;
private volatile long count = 0; // 计数器用于跟踪已生成的数据条数
private final Random random = new Random();
public void run(SourceContext<String> ctx) throws Exception {
while (running) {
int randomNum = random.nextInt(5) + 1;
long timestamp = System.currentTimeMillis();
ctx.collectWithTimestamp("key" + randomNum + "," + 1 + "," + timestamp, timestamp);
ZonedDateTime generateDataDateTime = Instant.ofEpochMilli(timestamp).atZone(ZoneId.systemDefault());
DateTimeFormatter formatter = DateTimeFormatter.ofPattern("yyyy-MM-dd HH:mm:ss.SSS");
String formattedGenerateDataDateTime = generateDataDateTime.format(formatter);
System.out.println("Generated data: " + "key" + randomNum + "," + 1 + "," + timestamp + " at " + formattedGenerateDataDateTime);
Thread.sleep(1000);
}
}
public void cancel() {
running = false;
}
});
DataStream<Tuple3<String, Integer, Long>> tuplesWithTimestamp = text.map(new MapFunction<String, Tuple3<String, Integer, Long>>() {
public Tuple3<String, Integer, Long> map(String value) {
String[] words = value.split(",");
return new Tuple3<>(words[0], Integer.parseInt(words[1]), Long.parseLong(words[2]));
}
}).returns(Types.TUPLE(Types.STRING, Types.INT, Types.LONG));
// 设置 Punctuated Watermark 策略
DataStream<Tuple3<String, Integer, Long>> withWatermarks = tuplesWithTimestamp.assignTimestampsAndWatermarks(new WatermarkStrategy<Tuple3<String, Integer, Long>>() {
public WatermarkGenerator<Tuple3<String, Integer, Long>> createWatermarkGenerator(WatermarkGeneratorSupplier.Context context) {
return new PunctuatedWatermarkGenerator();
}
public TimestampAssigner<Tuple3<String, Integer, Long>> createTimestampAssigner(TimestampAssignerSupplier.Context context) {
return new PunctuatedWatermarkGenerator();
}
});
// 窗口逻辑
DataStream<Tuple2<String, Integer>> keyedStream = withWatermarks
.keyBy(value -> value.f0)
.window(TumblingEventTimeWindows.of(Time.seconds(5)))
.process(new ProcessWindowFunction<Tuple3<String, Integer, Long>, Tuple2<String, Integer>, String, TimeWindow>() {
public void process(String key, Context context, Iterable<Tuple3<String, Integer, Long>> elements, Collector<Tuple2<String, Integer>> out) throws Exception {
int count = 0;
for (Tuple3<String, Integer, Long> element : elements) {
count++;
}
long start = context.window().getStart();
long end = context.window().getEnd();
ZonedDateTime startDateTime = Instant.ofEpochMilli(start).atZone(ZoneId.systemDefault());
ZonedDateTime endDateTime = Instant.ofEpochMilli(end).atZone(ZoneId.systemDefault());
DateTimeFormatter formatter = DateTimeFormatter.ofPattern("yyyy-MM-dd HH:mm:ss.SSS");
String formattedStart = startDateTime.format(formatter);
String formattedEnd = endDateTime.format(formatter);
System.out.println("Tumbling Window [start " + formattedStart + ", end " + formattedEnd + ") for key " + key);
// 输出窗口结束时的Watermark
long windowEndWatermark = context.currentWatermark();
ZonedDateTime windowEndDateTime = Instant.ofEpochMilli(windowEndWatermark).atZone(ZoneId.systemDefault());
String formattedWindowEndWatermark = windowEndDateTime.format(formatter);
System.out.println("Watermark at the end of window: " + formattedWindowEndWatermark);
out.collect(new Tuple2<>(key, count));
}
});
// 输出结果
keyedStream.print();
// 执行任务
env.execute("Punctuated Watermark Demo");
}
private static class PunctuatedWatermarkGenerator
implements WatermarkGenerator<Tuple3<String, Integer, Long>>, TimestampAssigner<Tuple3<String, Integer, Long>> {
private long maxTimestamp = Long.MIN_VALUE;
public long extractTimestamp(Tuple3<String, Integer, Long> element, long recordTimestamp) {
// 提前事件时间要先判断时间戳字段是否为-1
if (element.f2 != -1) {
return element.f2;
} else {
// 如果为空,返回上一次的事件时间
return recordTimestamp > 0 ? recordTimestamp : 0;
}
}
public void onEvent(Tuple3<String, Integer, Long> event, long eventTimestamp, WatermarkOutput output) {
maxTimestamp = Math.max(maxTimestamp, eventTimestamp);
if (event.f0.equals("key2")) {
System.out.println("Event: " + event.f0 + "," + event.f1 + "," + event.f2);
ZonedDateTime watermarkDateTime = Instant.ofEpochMilli(maxTimestamp).atZone(ZoneId.systemDefault());
DateTimeFormatter formatter = DateTimeFormatter.ofPattern("yyyy-MM-dd HH:mm:ss.SSS");
String formattedWatermark = watermarkDateTime.format(formatter);
System.out.println("Emitting Watermark: " + formattedWatermark);
output.emitWatermark(new Watermark(event.f2));
}
}
public void onPeriodicEmit(WatermarkOutput output) {
// nothing
}
}
}
这或许是一个对你有用的开源项目,data-warehouse-learning 项目是一套基于 MySQL + Kafka + Hadoop + Hive + Dolphinscheduler + Doris + Seatunnel + Paimon + Hudi + Iceberg + Flink + Dinky + DataRT + SuperSet 实现的实时离线数仓(数据湖)系统,以大家最熟悉的电商业务为切入点,详细讲述并实现了数据产生、同步、数据建模、数仓(数据湖)建设、数据服务、BI报表展示等数据全链路处理流程。
https://gitee.com/wzylzjtn/data-warehouse-learning
https://github.com/Mrkuhuo/data-warehouse-learning
https://bigdatacircle.top/
项目演示:
01
代码获取
https://gitee.com/wzylzjtn/data-warehouse-learning
https://github.com/Mrkuhuo/data-warehouse-learning
02
文档获取
03
进交流群群添加作者
推荐阅读
【视频】| 2024最新版【实时数仓数据湖实战教程视频】重磅更新,打通新手到大牛的最后一公里 【视频】| 1. Flink基本架构概述 【视频】| 2. Flink任务运行时架构图详解 【视频】| 3. Flink 任务、子任务、算子链及Slot共享详解 【视频】| 4. Flink四种API详解及代码演示 【视频】| 5. 如何在IDEA中启动Flink任务并看到WEBUI呢? 【视频】| 6. Centos8 安装Flink1.18.1 详细教程 【视频】| 7. Flink 使用 Session & Application 模式提交任务实战演示 【视频】| 8. Flink 三种时间语义详解及代码实战 【视频】| 9. Flink 窗口函数是个好东西,你真的会用吗? 【视频】| 10. Flink乱序问题解决神器Watermark详解及代码实战 【视频】| 11. Flink Watermark何种情况下可以触发窗口计算?