【实战】Spring Cloud Stream 3.1+整合Kafka

前言

之前我们已经整合过Spring Cloud Stream 3.0版本与Kafka、RabbitMQ中间件，简直不要太好，直接让我们不用再关心底层MQ如何集与消息收发。但是从Spring Cloud 2020版本开始，Spring Cloud Stream的版本升级至3.1.0以上版本，自此版本开始@StreamListener上面就增加@Deprecated注解，不赞成使用，有可能接下来的版本会删除掉。传说是有利于使用Project Reactor提供的事件流抽象（如Flux和Mono），命令函数在每个单独的事件上触发，而reactive函数只触发一次。故今天我们分享一期Spring Cloud Stream 3.1+整合Kafka，各位看官敬请鉴赏。

新版版本优势

新版提倡用函数式进行发送和消费信息

定义返回类型为Supplier, Function or Consumer的bean提供消息发送和消费的bean 看看绑定名称命名规则
input - + -in- +
output - + -out- +

在配置文件中指定spring.cloud.function.definition/spring.cloud.stream.function.definition的名称后会把这个bean绑定到对应的消费者和提供者上。

比如 inputChannel bean绑定了inputChannel-in-0通道,outputChannel bean绑定了outputChannel-out-0通道：

spring:kafka:bootstrap-servers: 192.168.112.10:9092,192.168.112.130:9092,192.168.112.129:9092cloud:stream:kafka:binder:brokers: ${spring.kafka.bootstrap-servers}binders:kafkahub:type: kafkaenvironment:spring:cloud:stream:kafka: ${spring.cloud.stream.kafka.binder}default-binder: kafkahub function:definition: inputChannel,outputChannelbindings:inputChannel-in-0:binder: kafkahubdestination: test-kafka-topicgroup: test-kafka-groupcontent-type: text/plainoutputChannel-out-0:binder: kafkahubdestination: test-kafka-topiccontent-type: text/plainproducer:partition-count: 3 #分区数目

此时消息生产者为：

@Resource
private StreamBridge streamBridge;@GetMapping("/send")
public Boolean sendMessageToKafka(String msg){boolean send = streamBridge.send("outputChannel-out-0", MessageBuilder.withPayload("kafka测试："+msg).build());return send;
}

此时消息消费者为：

@Configuration
public class KafkaChannel {@Resourceprivate StreamBridge streamBridge;/*** inputChannel 消费者* @author senfel* @date 2024/6/18 15:26* @return java.util.function.Consumer<java.lang.String>*/@Beanpublic Consumer<Message<String>> inputChannel(){return message -> {System.out.println("接收到消息Payloa：" + message.getPayload());System.out.println("接收到消息Header：" + message.getHeaders());};}

}

实战演示

我们简单进行一下演示即可，kafka环境可以看我之前的博文搭建。
主要演示功能：

正常情况下生产者发送消息到kafka,消费者监听消息并消费成功
异常情况下消费者消费失败，立即将异常消息投递到另一个topic上，兜底topic消费者消费

本次全部采用自动ack模式，如果需要手动ack参照之前的博文配置即可，注意在消费者端加上手动ack逻辑。

增加maven依赖

 <parent><groupId>org.springframework.boot</groupId><artifactId>spring-boot-starter-parent</artifactId><version>2.3.12.RELEASE</version><relativePath/> <!-- lookup parent from repository -->
</parent>
<groupId>com.example</groupId>
<artifactId>cce-demo</artifactId>
<version>0.0.1-SNAPSHOT</version>
<name>seata-demo-order</name>
<description>Demo project for Spring Boot</description>
<properties><java.version>8</java.version><spring-cloud.version>Hoxton.SR12</spring-cloud.version>
</properties>
<dependencies><dependency><groupId>org.springframework.cloud</groupId><artifactId>spring-cloud-starter-stream-kafka</artifactId><version>3.2.4</version></dependency>
</dependencies>
<dependencyManagement><dependencies><dependency><groupId>org.springframework.cloud</groupId><artifactId>spring-cloud-dependencies</artifactId><version>${spring-cloud.version}</version><type>pom</type><scope>import</scope></dependency></dependencies>
</dependencyManagement>

增加applicaiton.yaml配置

spring:#kafkakafka:bootstrap-servers: 192.168.112.10:9092,192.168.112.130:9092,192.168.112.129:9092cloud:stream:kafka:  # kafka配置binder:brokers: ${spring.kafka.bootstrap-servers}auto-add-partitions: true #自动分区auto-create-topics: true #自动创建主题replication-factor: 3 #副本min-partition-count: 3 #最小分区bindings:outputChannel-out-0:producer:# 无限制重发不产生消息丢失retries: Integer.MAX_VALUE#acks =0：producer不等待broker的ack，broker一接收到还没有写入磁盘就已经返回，可靠性最低#acks =1：producer等待broker的ack，partition的leader刷盘成功后返回ack，如果在follower同步成功之前leader故障，那么将会丢失数据，可靠性中#acks = all 、 -1：producer等待broker的ack，partition的leader和follower全部落盘成功后才返回ack，可靠性高,但延迟时间长#可以设置的值为：all, -1, 0, 1acks: allmin:insync:replicas: 3 #感知副本数inputChannel-in-0:consumer:concurrency: 1 #消费者数量max-concurrency: 5 #最大消费者数量recovery-interval: 3000  #3s 重连auto-rebalance-enabled: true  #主题分区消费者组成员自动平衡auto-commit-offset: false   #手动提交偏移量enable-dlq: true  # 开启 dlq队列dlq-name: test-kafka-topic.dlqdeserializationExceptionHandler: sendToDlq #异常加入死信binders: # 与外部mq组件绑定kafkahub:type: kafkaenvironment:spring:cloud:stream:kafka: ${spring.cloud.stream.kafka.binder}default-binder: kafkahub #默认绑定function: # 定义channel名字，每个channel又可以作为生产者（in）与消费者（out）definition: inputChannel;outputChannel;dlqChannelbindings: # 通道绑定inputChannel-in-0:binder: kafkahubdestination: test-kafka-topicgroup: test-kafka-groupcontent-type: text/plainconsumer:maxAttempts: 1 # 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3backOffInitialInterval: 1000 # 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2backOffMaxInterval: 10000 # 下一次尝试重试的最大时间间隔，默认为10000ms，即10soutputChannel-out-0:binder: kafkahubdestination: test-kafka-topiccontent-type: text/plainproducer:partition-count: 3 #分区数目dlqChannel-in-0:binder: kafkahubdestination: test-kafka-topic.dlqgroup: test-kafka-groupcontent-type: text/plainconsumer:maxAttempts: 1 # 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3backOffInitialInterval: 1000 # 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2backOffMaxInterval: 10000 # 下一次尝试重试的最大时间间隔，默认为10000ms，即10sdlqChannel-out-0:binder: kafkahubdestination: test-kafka-topic.dlqcontent-type: text/plainproducer:partition-count: 3 #分区数目

新增Kafka通道消费者

import org.springframework.cloud.stream.function.StreamBridge;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import org.springframework.messaging.Message;
import org.springframework.messaging.support.MessageBuilder;
import javax.annotation.Resource;
import java.util.function.Consumer;/*** KafkaCustomer* @author senfel* @version 1.0* @date 2024/6/18 15:22*/
@Configuration
public class KafkaChannel {@Resourceprivate StreamBridge streamBridge;/*** inputChannel 消费者* @author senfel* @date 2024/6/18 15:26* @return java.util.function.Consumer<java.lang.String>*/@Beanpublic Consumer<Message<String>> inputChannel(){return message -> {System.out.println("接收到消息：" + message.getPayload());System.out.println("接收到消息：" + message.getHeaders());if(message.getPayload().contains("9")){boolean send = streamBridge.send("dlqChannel-out-0", MessageBuilder.withPayload("kafka异常消息发送到dlq测试："+message).build());System.err.println("向dlqChannel发送消息："+send);}};}/*** dlqChannel 死信消费者* @author senfel* @date 2024/6/18 15:26* @return java.util.function.Consumer<java.lang.String>*/@Beanpublic Consumer<Message<String>> dlqChannel(){return message -> {System.out.println("死信dlqChannel接收到消息：" + message.getPayload());System.out.println("死信dlqChannel接收到消息：" + message.getHeaders());};}
}

新增发送消息的接口

@Resource
private StreamBridge streamBridge;@GetMapping("/send")
public Boolean sendMessageToKafka(String msg){boolean send = streamBridge.send("outputChannel-out-0", MessageBuilder.withPayload("kafka测试："+msg).build());return send;
}

实战测试

postman发送一个正常的消息

postman发送异常消息