春靴Kafka客户端有“断路器”吗？

Question

如果Kafka服务器（暂时）关闭，我的Spring Boot应用程序reactiveKafKaconsumerTemplate将继续尝试连接，但失败，从而导致不必要的通信量并扰乱日志文件：

2021-11-10 14:45:30.265  WARN 24984 --- [onsumer-group-1] org.apache.kafka.clients.NetworkClient   : [Consumer clientId=consumer-group-1, groupId=consumer-group] Connection to node -1 (localhost/127.0.0.1:29092) could not be established. Broker may not be available.
2021-11-10 14:45:32.792  WARN 24984 --- [onsumer-group-1] org.apache.kafka.clients.NetworkClient   : [Consumer clientId=consumer-group-1, groupId=consumer-group] Bootstrap broker localhost:29092 (id: -1 rack: null) disconnected
2021-11-10 14:45:34.845  WARN 24984 --- [onsumer-group-1] org.apache.kafka.clients.NetworkClient   : [Consumer clientId=consumer-group-1, groupId=consumer-group] Connection to node -1 (localhost/127.0.0.1:29092) could not be established. Broker may not be available.
2021-11-10 14:45:34.845  WARN 24984 --- [onsumer-group-1] org.apache.kafka.clients.NetworkClient   : [Consumer clientId=consumer-group-1, groupId=consumer-group] Bootstrap broker localhost:29092 (id: -1 rack: null) disconnected

我知道参数reconnect.backoff.ms，这就是我如何创建reactiveKafKaconsumerTemplatebean：

@Bean
public ReactiveKafkaConsumerTemplate<String, MyEvent> kafkaConsumer(KafkaProperties properties) {
    final Map<String, Object> map = new HashMap<>(properties.buildConsumerProperties());
    map.put(ConsumerConfig.GROUP_ID_CONFIG, "MyGroup");
    map.put(ConsumerConfig.RECONNECT_BACKOFF_MS_CONFIG, 10_000L);
    final JsonDeserializer<DisplayCurrencyEvent> jsonDeserializer = new JsonDeserializer<>();
    jsonDeserializer.addTrustedPackages("com.example.myapplication");

    return new ReactiveKafkaConsumerTemplate<>(
            ReceiverOptions
                    .<String, MyEvent>create(map)
                    .withKeyDeserializer(new ErrorHandlingDeserializer<>(new StringDeserializer()))
                    .withValueDeserializer(new ErrorHandlingDeserializer<>(jsonDeserializer))
                    .subscription(List.of("MyTopic")));
}

消费者仍然试图每3秒连接一次。

Answer 1

参见https://kafka.apache.org/documentation/#consumerconfigs_retry.backoff.ms

试图重新连接到给定主机之前等待的基本时间。这避免了在紧密循环中重复连接到主机。此退避适用于客户端到代理的所有连接尝试。

和https://kafka.apache.org/documentation/#consumerconfigs_reconnect.backoff.max.ms

重新连接到多次连接失败的代理时等待的最长时间（毫秒）。如果提供，则每台主机的退避量将随着连续的连接失败而呈指数增长，直至此最大值。计算退避增加后，增加20%的随机抖动以避免连接Storm。

和