flink读写kafka--读kafka

介绍

主要介绍实际中flink如何读取写入设置kafka

flink版本：1.13.2

github地址：https://github.com/dahai1996/mdw-flink-quickstart

读取kafka

引入依赖

    
        org.apache.flink
        flink-connector-kafka_2.11
        ${flink.version}
    

设置并创建kafka source

        Properties propertiesSource = new Properties();
        propertiesSource.setProperty(ConsumerConfig.BOOTSTRAP_SERVERS_CONFIG, "ip");
        propertiesSource.setProperty(ConsumerConfig.GROUP_ID_CONFIG, groupId);
        propertiesSource.setProperty(ConsumerConfig.REQUEST_TIMEOUT_MS_CONFIG,"60000");
        propertiesSource.setProperty(ConsumerConfig.AUTO_OFFSET_RESET_CONFIG,"earliest");

        FlinkKafkaConsumer source = new FlinkKafkaConsumer<>(topic, new SimpleStringSchema(), propertiesSource);

注：SimpleStringSchema 是序列化方法，简单地读取为string，一般都使用这个

用建造者模式包装下

public class SourceKafkaBuilder {
    private final Properties properties = new Properties();
    private FlinkKafkaConsumer sourceKafka;
    private final String topic;
    private final DeserializationSchema valueDeserializer;

    public SourceKafkaBuilder(RunEnv runEnv,String topic,String groupId,DeserializationSchema valueDeserializer) {
        properties.setProperty(ConsumerConfig.BOOTSTRAP_SERVERS_CONFIG, runEnv.getKafkaHost());
        properties.setProperty(ConsumerConfig.GROUP_ID_CONFIG, groupId);

        this.topic=topic;
        this.valueDeserializer=valueDeserializer;
    }

    public SourceKafkaBuilder setSessionTimeOutMs(String sessionTimeOutMs){
        properties.setProperty(ConsumerConfig.SESSION_TIMEOUT_MS_CONFIG, sessionTimeOutMs);
        return this;
    }

    public SourceKafkaBuilder setRequestTimeOutMs(String requestTimeOutMs){
        properties.setProperty(ConsumerConfig.REQUEST_TIMEOUT_MS_CONFIG, requestTimeOutMs);
        return this;
    }
    public SourceKafkaBuilder setAutoOffsetResetConfig(String autoOffsetResetConfig){
        properties.setProperty(ConsumerConfig.AUTO_OFFSET_RESET_CONFIG, autoOffsetResetConfig);
        return this;
    }
    public SourceKafkaBuilder setExactlyOnce(){
        properties.setProperty(ConsumerConfig.ISOLATION_LEVEL_CONFIG, "read_committed");
        return this;
    }

    /**
     * @param key ConsumerConfig中包含的key
     * @see ConsumerConfig
     * @param value 值
     * @return 构造器
     */
    public SourceKafkaBuilder setPropertyValue(String key,String value){
        properties.setProperty(key, value);
        return this;
    }

    public FlinkKafkaConsumer build(){
        return new FlinkKafkaConsumer<>(topic,valueDeserializer, properties);
    }
}

使用：

FlinkKafkaConsumerBase sourceKafka = new SourceKafkaBuilder<>(uat, topicName, groupId, new SimpleStringSchema())
                .setRequestTimeOutMs("60000")
                .setSessionTimeOutMs("60000")
                .build()
                .setStartFromGroupOffsets();

注：更多设置使用setPropertyValue（）方法，具体有哪些参数，参考ConsumerConfig类的常量。

注2：setExactlyOnce 方法与端到端一致性有关，后续会提到