Spring-Kafka-如何使用注释为侦听器指定分区范围?

时间:2018-12-03 06:38:17

标签: spring-kafka

我有3个主题,每个主题都有50个分区。我想使用@KafkaListener为不同的分区指定侦听器

  • 每个主题都应该有一个听众吗?

  • 我应该为一个主题设置多个侦听器,如果可以的话,如何为该主题指定分区范围?

  • 另外,在这3个主题中,有2个主题的数据要比第3个主题多得多,所以我应该为此主题吸引更多的听众吗?

  • 我应该如何选择并发号码?

到目前为止,我的代码:

    @Configuration
    @EnableKafka
public class ConsumerConfig {

  // Factory to create the consumer classes
  @Bean
  public ConsumerFactory<String, String> consumerFactory() {
    return new DefaultKafkaConsumerFactory<String, String>(consumerConfigs());
  }

  @Bean
  public Map<String, Object> consumerConfigs() {
    Map<String, Object> propsMap = new HashMap<>();
    propsMap.put(ConsumerConfig.BOOTSTRAP_SERVERS_CONFIG, "localhost:9092");
    propsMap.put(ConsumerConfig.ENABLE_AUTO_COMMIT_CONFIG, false);
    propsMap.put(ConsumerConfig.AUTO_COMMIT_INTERVAL_MS_CONFIG, "100");
    propsMap.put(ConsumerConfig.SESSION_TIMEOUT_MS_CONFIG, "15000");
    propsMap.put(ConsumerConfig.KEY_DESERIALIZER_CLASS_CONFIG, StringDeserializer.class);
    propsMap.put(ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG, StringDeserializer.class);
    propsMap.put(ConsumerConfig.GROUP_ID_CONFIG, "group1");
    propsMap.put(ConsumerConfig.AUTO_OFFSET_RESET_CONFIG, "earliest");
    return propsMap;
  }


  //
  @Bean
  KafkaListenerContainerFactory<ConcurrentMessageListenerContainer<String, String>> kafkaListenerContainerFactory() {
    ConcurrentKafkaListenerContainerFactory<String, String> factory = new ConcurrentKafkaListenerContainerFactory<>();
    factory.setConsumerFactory(consumerFactory());
    factory.setConcurrency(3);
    factory.getContainerProperties().setPollTimeout(3000);
    return factory;
  }

  @Bean
  public Listener listener() {
    return new Listener();
  }
}

1 个答案:

答案 0 :(得分:1)

请参见the documentation

  

您还可以为POJO侦听器配置明确的主题和分区(以及可选的初始偏移量):

@KafkaListener(id = "bar", topicPartitions =
        { @TopicPartition(topic = "topic1", partitions = { "0", "1" }),
          @TopicPartition(topic = "topic2", partitions = "0",
             partitionOffsets = @PartitionOffset(partition = "1", initialOffset = "100"))
        })
public void listen(ConsumerRecord<?, ?> record) {
    ...
}

一个侦听器可以侦听多个主题,但是如果您的消息量不同,我建议为每个主题分配一个单独的侦听器;否则,小批量主题可能无法获得所需的活动。

编辑

您可以使用SpEL表达式生成分区数组。

例如;两个侦听器,一个配置为奇数分区,另一个配置为偶数……

@SpringBootApplication
public class So53588657Application {

    public static void main(String[] args) {
        SpringApplication.run(So53588657Application.class, args);
    }

    @Bean
    public NewTopic topic() {
        return new NewTopic("so53588657", 50, (short) 1);
    }

    @KafkaListener(id = "odd", topicPartitions =
            @TopicPartition(topic = "so53588657",
                partitions = "#{T(com.example.So53588657Application$SplitParts).odds(50)}"))
    public void oddParts(String in) {
        // ...
    }

    @KafkaListener(id = "even", topicPartitions =
            @TopicPartition(topic = "so53588657",
                partitions = "#{T(com.example.So53588657Application$SplitParts).evens(50)}"))
    public void evenParts(String in) {
        // ...
    }

    @Bean
    public ApplicationRunner runner(KafkaListenerEndpointRegistry registry) {
        return args -> registry.getListenerContainers()
                .forEach(c -> c.getAssignedPartitions().forEach(ap -> System.out.println(ap)));
    }

    public static class SplitParts {

        public static String[] odds(int partitions) {
            return split(partitions, i -> i % 2 == 0);
        }

        public static String[] evens(int partitions) {
            return split(partitions, i -> i % 2 == 1);
        }

        private static String[] split(int partitions, IntPredicate predicate) {
            return IntStream.range(0, partitions)
                    .filter(predicate)
                    .mapToObj(i -> String.valueOf(i))
                    .collect(Collectors.toList())
                    .toArray(new String[0]);
        }

    }

}

或者您可以在属性中以逗号分隔列表的形式提供它们并使用

partitions = { "#{'${partition.list}'.split(',')}" })