Spring Kafka 消息重試和死信隊列

Spring-kafka內部封裝了可重試消費消息的語義，也就是可以設置爲當消費數據出現異常時，重試這個消息。而且可以設置重試達到多少次後，讓消息進入預定好的Topic。也就是死信隊列裏。

demo

@Component
@EnableScheduling
public class DemoListener {

    private static final Logger logger = LoggerFactory.getLogger(DemoListener.class);

    @Autowired
    private ConsumerFactory consumerFactory;

    @Autowired
    private KafkaTemplate kafkaTemplate;

    @Bean
    public ConcurrentKafkaListenerContainerFactory containerFactory() {
        ConcurrentKafkaListenerContainerFactory factory = new ConcurrentKafkaListenerContainerFactory();
        factory.setConsumerFactory(consumerFactory);
        // 最大重試次數3次
        factory.setErrorHandler(new SeekToCurrentErrorHandler(new DeadLetterPublishingRecoverer(kafkaTemplate), 3));
        return factory;
    }

    @KafkaListener(id = "group", topics = "testTopic", containerFactory = "containerFactory")
    public void listen(ConsumerRecord record, Acknowledgment ack) throws Exception {
        Optional kafkaMessage = Optional.ofNullable(record.value());

        if (!kafkaMessage.isPresent()) {
            throw new Exception("監聽到的消息爲空值");
        }

        logger.info("topicID: " + record.topic());
        logger.info("recordValue: " + record.value());

        try {
            /*業務邏輯*/
            ack.acknowledge();
        } catch (Exception e) {
            throw new Exception("消息異常，進入死信隊列...");
        }
    }


    @KafkaListener(id = "dltGroup", topics = "testTopic.DLT")
    public void dltListen(String input) {
        logger.info("Received from DLT: " + input);
    }

}

demo中，監聽testTopic主題的監聽器在處理業務邏輯時如果觸發運行時異常，監聽器會重新嘗試三次調用，當到達最大的重試次數後。消息就會被丟掉重試死信隊列裏面去。死信隊列的Topic的規則是，業務Topic名字+“.DLT”。如上面業務Topic的name爲“testTopic”，那麼對應的死信隊列的Topic就是“testTopic.DLT”