11

我可以向社区查询一下收听多个主题的最佳方式是什么，每个主题都包含不同类别的消息？

在过去的几天里，我一直在玩 Spring Kafka。到目前为止我的思考过程：

  • 因为在初始化 KafkaListenerContainerFactory 时需要将反序列化器传递给 DefaultKafkaConsumerFactory。这似乎表明，如果我需要多个容器，每个容器都反序列化不同类型的消息，我将无法使用 @EnableKafka 和 @KafkaListener 注释。

  • 这使我认为这样做的唯一方法是实例化多个 KafkaMessageListenerContainers。

  • 鉴于 KafkaMessageListenerContainers 是单线程的，我需要同时收听多个主题，我真的应该使用多个 ConcurrentKafkaMessageListenerContainers。

我会在正确的轨道上吗？有一个更好的方法吗？

谢谢！

4

3 回答 3

14

这是一个非常简单的例子。

// -----------------------------------------------
// Sender
// -----------------------------------------------

@Configuration
public class SenderConfig {
    @Bean
    public Map<String, Object> producerConfigs() {
        Map<String, Object> props = new HashMap<>();
        ......
        props.put(ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG, StringSerializer.class);
        props.put(ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG, JsonSerializer.class);
        return props;
    }

    @Bean
    public ProducerFactory<String, Class1> producerFactory1() {
        return new DefaultKafkaProducerFactory<String, Class1>(producerConfigs());
    }

    @Bean
    public KafkaTemplate<String, Class1> kafkaTemplate1() {
        return new KafkaTemplate<>(producerFactory1());
    }

    @Bean
    public Sender1 sender1() {
        return new Sender1();
    }

    //-------- send the second class --------

    @Bean
    public ProducerFactory<String, Class2> producerFactory2() {
        return new DefaultKafkaProducerFactory<String, Class2>(producerConfigs());
    }

    @Bean
    public KafkaTemplate<String, Class2> kafkaTemplate2() {
        return new KafkaTemplate<>(producerFactory2());
    }

    @Bean
    public Sender2 sender2() {
        return new Sender2();
    }
}

public class Sender1 {
    @Autowired
    private KafkaTemplate<String, Class1> kafkaTemplate1;

    public void send(String topic, Class1 c1) {
        kafkaTemplate1.send(topic, c1);
   }
}

public class Sender2 {
    @Autowired
    private KafkaTemplate<String, Class2> kafkaTemplate2;

    public void send(String topic, Class2 c2) {
        kafkaTemplate2.send(topic, c2);
    }
}

// -----------------------------------------------
// Receiver
// -----------------------------------------------

@Configuration
@EnableKafka
public class ReceiverConfig {

    @Bean
    public Map<String, Object> consumerConfigs() {
        Map<String, Object> props = new HashMap<>();
        ......
        props.put(ConsumerConfig.KEY_DESERIALIZER_CLASS_CONFIG, StringDeserializer.class);
        props.put(ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG, JsonDeserializer.class);
        return props;
    }

    @Bean
    public ConsumerFactory<String, Class1> consumerFactory1() {
        return new DefaultKafkaConsumerFactory<>(consumerConfigs(), new StringDeserializer(),
            new JsonDeserializer<>(Class1.class));
    }

    @Bean
    public ConcurrentKafkaListenerContainerFactory<String, Class1> kafkaListenerContainerFactory1() {
        ConcurrentKafkaListenerContainerFactory<String, Class1> factory =
            new ConcurrentKafkaListenerContainerFactory<>();
    factory.setConsumerFactory(consumerFactory1());
        return factory;
    }

    @Bean
    public Receiver1 receiver1() {
        return new Receiver1();
    }

    //-------- add the second listener

    @Bean
    public ConsumerFactory<String, Class2> consumerFactory2() {
    return new DefaultKafkaConsumerFactory<>(consumerConfigs(), new StringDeserializer(),
            new JsonDeserializer<>(Class2.class));
    }

    @Bean
    public ConcurrentKafkaListenerContainerFactory<String, Class2> kafkaListenerContainerFactory2() {
        ConcurrentKafkaListenerContainerFactory<String, Class2> factory =
            new ConcurrentKafkaListenerContainerFactory<>();
        factory.setConsumerFactory(consumerFactory2());
        return factory;
    }

    @Bean
    public Receiver2 receiver2() {
        return new Receiver2();
    }
}

public class Receiver1 {
    @KafkaListener(id="listener1", topics = "topic1", containerFactory = "kafkaListenerContainerFactory1")
    public void receive(Class1 c1) {
        LOGGER.info("Received c1");
    }
}

public class Receiver2 {
    @KafkaListener(id="listener2", topics = "topic2", containerFactory = "kafkaListenerContainerFactory2")
    public void receive(Class2 c2) {
        LOGGER.info("Received c2");
    }
}
于 2017-10-10T15:45:09.773 回答
7

您可以使用注释，您只需要为每个使用不同的侦听器容器工厂。

该框架将为每个注释创建一个侦听器容器。

您还可以在单​​线程容器上收听多个主题，但它们将在单个线程上进行处理。

看看我去年 SpringOne 平台演讲中的代码——你可能想看看 app6，它展示了如何使用 aMessageConverter而不是反序列化器，这可能有助于简化你的配置。

于 2017-03-31T13:11:16.157 回答
3

我想用下面的代码来应用你的感觉

@Configuration
@EnableKafka
public class ConsumerConfig {

    @Value("${kafka.bootstrap-servers}")
    private String bootstrapServers;

    @Value("${kafka.group-id}")
    private String groupId;

    /**
     * Configuration of Consumer properties.
     * 
     * @return
     */
    //@Bean
    public Map<String, Object> consumerConfigs() {
        Map<String, Object> props = new HashMap<>();

        props.put(ConsumerConfig.BOOTSTRAP_SERVERS_CONFIG, bootstrapServers);
        props.put(ConsumerConfig.KEY_DESERIALIZER_CLASS_CONFIG, StringDeserializer.class);
        props.put(ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG, JsonDeserializer.class);

        props.put(ConsumerConfig.GROUP_ID_CONFIG, groupId);

        return props;
    }

    //@Bean
    public ConsumerFactory<String, ClassA> consumerFactory1() {
        return new DefaultKafkaConsumerFactory<>(consumerConfigs(), new StringDeserializer(),
                new ClassA());
    }

    /**
     * Kafka Listener Container Factory.
     * @return
     */
    @Bean("kafkaListenerContainerFactory1")
    public KafkaListenerContainerFactory<ConcurrentMessageListenerContainer<String, ClassA>> kafkaListenerContainerFactory1() {
        ConcurrentKafkaListenerContainerFactory<String, ClassA> factory;
        factory = new ConcurrentKafkaListenerContainerFactory<>();
        factory.setConsumerFactory(consumerFactory1());
        return factory;
    }

    //@Bean
    public ConsumerFactory<String, ClassB> consumerFactory2() {
        return new DefaultKafkaConsumerFactory<>(consumerConfigs(), new StringDeserializer(),
                new ClassB());
    }

    /**
     * Kafka Listener Container Factory.
     * @return
     */
    @Bean("kafkaListenerContainerFactory2")
    public KafkaListenerContainerFactory<ConcurrentMessageListenerContainer<String, ClassB>> kafkaListenerContainerFactory2() {
        ConcurrentKafkaListenerContainerFactory<String, ClassB> factory;
        factory = new ConcurrentKafkaListenerContainerFactory<>();
        factory.setConsumerFactory(consumerFactory2());
        return factory;
    }

    @Bean
    public ReceiverClass receiver() {
        return new ReceiverClass();
    }

    class ReceiverClass {
        @KafkaListener(topics = "topic1", group = "group-id-test",
                containerFactory = "kafkaListenerContainerFactory1")
        public void receiveTopic1(ClassA a) {
            System.out.println("ReceiverClass.receive() ClassA : " + a);
        }

        @KafkaListener(topics = "topic2", group = "group-id-test",
                containerFactory = "kafkaListenerContainerFactory2")
        public void receiveTopic2(ClassB b) {
            System.out.println("ReceiverClass.receive() Classb : " + b);
        }

    }

    class ClassB  implements Deserializer  {

        @Override
        public void configure(Map configs, boolean isKey) {
            // TODO Auto-generated method stub

        }

        @Override
        public Object deserialize(String topic, byte[] data) {
            // TODO Auto-generated method stub
            return null;
        }

        @Override
        public void close() {
            // TODO Auto-generated method stub

        }

    }

    class ClassA  implements Deserializer {

        @Override
        public void configure(Map configs, boolean isKey) {
            // TODO Auto-generated method stub

        }

        @Override
        public Object deserialize(String topic, byte[] data) {
            // TODO Auto-generated method stub
            return null;
        }

        @Override
        public void close() {
            // TODO Auto-generated method stub

        }

    }
}
于 2018-09-10T17:44:52.397 回答