Question

我正在用Java实现Flink程序来使用MapStateDescriptor处理状态。我将基于此source进行实施。由于某些原因，MapState保留了先前的值，因此我无法计算所需的平均值。在调试averageTemp时始终为空，并且在其中找不到任何密钥。我在实现过程中缺少什么？

import java.util.Map;

import org.apache.flink.api.common.functions.RichMapFunction;
import org.apache.flink.api.common.state.MapState;
import org.apache.flink.api.common.state.MapStateDescriptor;
import org.apache.flink.api.java.functions.KeySelector;
import org.apache.flink.api.java.tuple.Tuple2;
import org.apache.flink.configuration.Configuration;
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.sense.flink.mqtt.MqttTemperature;
import org.sense.flink.mqtt.TemperatureMqttConsumer;

public class SensorsMultipleReadingMqttEdgentQEP {

    public SensorsMultipleReadingMqttEdgentQEP() throws Exception {
        StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
        env.setStreamTimeCharacteristic(TimeCharacteristic.IngestionTime);

        DataStream<MqttTemperature> temperatureStream01 = env.addSource(new TemperatureMqttConsumer("topic-edgent-01"));
        DataStream<MqttTemperature> temperatureStream02 = env.addSource(new TemperatureMqttConsumer("topic-edgent-02"));
        DataStream<MqttTemperature> temperatureStream03 = env.addSource(new TemperatureMqttConsumer("topic-edgent-03"));
        DataStream<MqttTemperature> temperatureStreams = temperatureStream01.union(temperatureStream02)
                .union(temperatureStream03);

        DataStream<Tuple2<String, Double>> average = temperatureStreams.keyBy(new TemperatureKeySelector())
                .map(new AverageTempMapper());
        average.print();

        env.execute("SensorsMultipleReadingMqttEdgentQEP");
    }

    public static class TemperatureKeySelector implements KeySelector<MqttTemperature, Integer> {

        private static final long serialVersionUID = 5905504239899133953L;

        @Override
        public Integer getKey(MqttTemperature value) throws Exception {
            return value.getId();
        }
    }

    public static class AverageTempMapper extends RichMapFunction<MqttTemperature, Tuple2<String, Double>> {

        private static final long serialVersionUID = -5489672634096634902L;
        private MapState<String, Double> averageTemp;

        @Override
        public void open(Configuration parameters) throws Exception {
            averageTemp = getRuntimeContext()
                    .getMapState(new MapStateDescriptor<>("average-temperature", String.class, Double.class));
        }

        @Override
        public Tuple2<String, Double> map(MqttTemperature value) throws Exception {
            String key = "no-room";
            Double temp = value.getTemp();

            if (value.getId().equals(1) || value.getId().equals(2) || value.getId().equals(3)) {
                key = "room-A";
            } else if (value.getId().equals(4) || value.getId().equals(5) || value.getId().equals(6)) {
                key = "room-B";
            } else if (value.getId().equals(7) || value.getId().equals(8) || value.getId().equals(9)) {
                key = "room-C";
            }
            // NEVER ITERATES ON THE averageTemp
            for (Map.Entry<String, Double> entry: averageTemp.entries()) {
                System.out.println(entry.getKey() + " - " + entry.getValue());
            }

            System.out.println("value: " + value);
            if (averageTemp.contains(key)) { // NEVER CONTAINS A KEY
                System.out.println("yes: " + key);
                temp = (averageTemp.get(key) + value.getTemp()) / 2;
            } else {
                averageTemp.put(key, temp);
            }
            return Tuple2.of(key, temp);
        }
    }
}

**编辑：**好。我误解了这个问题。该代码将先前的状态保存在MapState上。我错了，因为我正在调试代码。但是实际上我遇到的问题是，它启动了多个线程，并且在开始计算平均值之前，它至少覆盖了我地图的值三倍。

Answer 1

地图功能中的状态基于每个键。因此，当您调用map函数并获得map状态时，它将用于MqttTemperature记录中正在处理的id。

鉴于您想要每个房间的平均温度，我的处理方式如下：

根据id字段更改TemperatureKeySelector以返回room-A，room-B或room-C。
在AverageTempMapper中，有两个ValueState变量-一个是温度的总和（一个Double），另一个是一个计数。调用您的map()方法时，如果这两个ValueState变量之一为空，则将其初始化为0，然后求和/递增。

为什么我的Flink中的MapState变量不保留以前的值？

1 个答案: