SpringBoot整合SpringCloudStream3.1+版本的Kafka死信队列

上一篇直通车

SpringBoot整合SpringCloudStream3.1+版本Kafka

实现死信队列步骤

1. 添加死信队列配置文件，添加对应channel
2. 通道绑定配置对应的channel位置添加重试配置

结果

配置文件

Kafka基本配置（application-mq.yml）

server:
  port: 7105
spring:
  application:
	name: betrice-message-queue
  config:
	import:
	- classpath:application-bindings.yml
  cloud:
	stream:
	  kafka:
		binder:
		  brokers: localhost:9092
		  configuration:
			key-serializer: org.apache.kafka.common.serialization.StringSerializer
			value-serializer: org.apache.kafka.common.serialization.StringSerializer
		  consumer-properties:
			enable.auto.commit: false
	  binders:
		betrice-kafka:
		  type: kafka
		  environment:
			spring.kafka:
		  bootstrap-servers: ${spring.cloud.stream.kafka.binder.brokers}

创建死信队列配置文件（application-dql.yml）

spring:
  cloud:
	stream:
	  kafka:
		bindings:
		  dqlTransfer-in-0:
			consumer:
			  # When set to true, it enables DLQ behavior for the consumer. By default, messages that result in errors are forwarded to a topic named error.<destination>.<group>.
			  # messages sent to the DLQ topic are enhanced with the following headers: x-original-topic, x-exception-message, and x-exception-stacktrace as byte[].
			  # By default, a failed record is sent to the same partition number in the DLQ topic as the original record.
			  enableDlq: true
			  dlqName: Evad05-message-dlq
			  keySerde: org.apache.kafka.common.serialization.Serdes$StringSerde
#              valueSerde: org.apache.kafka.common.serialization.Serdes$StringSerde
			  valueSerde: com.devilvan.pojo.Evad05MessageSerde
			  autoCommitOnError: true
			  autoCommitOffset: true

注意：这里的valueSerde使用了对象类型，需要搭配application/json使用，consumer接收到消息后会转化为json字符串

通道绑定文件添加配置（application-bindings.yml）

channel对应上方配置文件的dqlTransfer-in-0

spring:
  cloud:
	stream:
	  betrice-default-binder: betrice-kafka
	  function:
		# 声明两个channel，transfer接收生产者的消息，处理完后给sink
		definition: transfer;sink;gather;gatherEcho;dqlTransfer;evad05DlqConsumer
	  bindings:
		# 添加生产者bindiing，输出到destination对应的topic
		dqlTransfer-in-0:
		  destination: Evad10
		  binder: ${spring.cloud.stream.betrice-default-binder}
		  group: evad05DlqConsumer # 使用死信队列必须要有group
		  content-type: application/json
		  consumer:
			maxAttempts: 2 # 当消息消费失败时，尝试消费该消息的最大次数（消息消费失败后，发布者会重新投递）。默认3
			backOffInitialInterval: 1000 # 消息消费失败后重试消费消息的初始化间隔时间。默认1s，即第一次重试消费会在1s后进行
			backOffMultiplier: 2 # 相邻两次重试之间的间隔时间的倍数。默认2，即第二次是第一次间隔时间的2倍，第三次是第二次的2倍
			backOffMaxInterval: 10000 # 下一次尝试重试的最大时间间隔，默认为10000ms，即10s。
		dqlTransfer-out-0:
		  destination: Evad10
		  binder: ${spring.cloud.stream.betrice-default-binder}
		  content-type: text/plain
		# 消费死信队列中的消息
		evad05DlqConsumer-in-0:
		  destination: Evad05-message-dlq
		  binder: ${spring.cloud.stream.betrice-default-binder}
		  content-type: text/plain

Controller

发送消息并将消息引入死信队列

@Slf4j
@RestController
@RequestMapping(value = "betriceMqController")
public class BetriceMqController {
	@Resource(name = "streamBridgeUtils")
	private StreamBridge streamBridge;

	@PostMapping("streamSend")
	public void streamSend(String topic, String message) {
		try {
			streamBridge.send(topic, message);
			log.info("发送消息：" + message);
		} catch (Exception e) {
			log.error("异常消息：" + e);
		}
	}

	@PostMapping("streamSendDql")
	public void streamSendDql(String topic, String message) {
		try {
			streamBridge.send(topic, message);
			log.info("发送消息：" + message);
		} catch (Exception e) {
			log.error("异常消息：" + e);
		}
	}

	@PostMapping("streamSendJsonDql")
	public void streamSendJsonDql(String topic) {
		try {
			Evad05MessageSerde message = new Evad05MessageSerde();
			message.setData("evad05 test dql");
			message.setCount(1);
			streamBridge.send(topic, message);
			log.info("发送消息：" + message);
		} catch (Exception e) {
			log.error("异常消息：" + e);
		}
	}
}

Channel

这里使用了transfer通道，消息从Evad10（topic）传来，经过transfer()方法后抛出异常，随后进入对应的死信队列

@Configuration
public class BetriceMqSubChannel {
	@Bean
	public Function<String, String> dqlTransfer() {
		return message -> {
			System.out.println("transfer: " + message);
			throw new RuntimeException("死信队列测试！");
		};
	}

	@Bean
	public Consumer<String> evad05DlqConsumer() {
		return message -> {
			System.out.println("Topic: evad05 Dlq Consumer: " + message);
		};
	}
}

将自定义序列化类型转换为JSON消息

步骤

1. 通道绑定文件（application-bindings.yml）的valueSerde属性添加自定义的序列化

2. BetriceMqController中封装该自定义类型的对象，并作为消息发送

@PostMapping("streamSendJsonDql")
public void streamSendJsonDql(String topic) {
	try {
		Evad05MessageSerde message = new Evad05MessageSerde();
		message.setData("evad05 test dql");
		message.setCount(1);
		streamBridge.send(topic, message);
		log.info("发送消息：" + message);
	} catch (Exception e) {
		log.error("异常消息：" + e);
	}
}

3. channel（BetriceMqSubChannel）接收到该消息并反序列化

@Bean
public Consumer<String> evad05DlqConsumer() {
	return message -> {
		System.out.println("Topic: evad05 Dlq Consumer: " + JSON.parseObject(message, Evad05MessageSerde.class));
	};
}

4. 结果

参考网址

Kafka 消费端消费重试和死信队列 - Java小强技术博客 (javacui.com)
spring cloud stream kafka rabbit 实现死信队列_spring cloud stream kafka 死信队列_it噩梦的博客-CSDN博客