RabbitMQ消费端并发和限流设置

并发

默认情况一下，一个listener对应一个consumer，如果想对应多个，有两种方式。

1. 在消费端，配置prefetch和concurrency参数便可以实现消费端MQ并发处理消息

   spring:
     rabbitmq:
       host: localhost
       port: 5672
       username: guest
       password: guest
       listener:
         simple:
   #        acknowledge-mode: manual  # 手动确定（默认自动确认）
           concurrency: 1 # 消费端的监听个数(即@RabbitListener开启几个线程去处理数据。)
           max-concurrency: 10 # 消费端的监听最大个数
           prefetch: 10
       connection-timeout: 15000   # 超时时间
   

   这个是个全局配置，应用里的任何队列对应的listener都至少有5个consumer，但是最好别这么做，因为一般情况下，一个listener对应一个consumer是够用的。只是针对部分场景，才需要一对多。

2. 在注解中配置该参数

   在RabbitListeners注解上加上concurrency属性

   会覆盖配置文件中的参数

   直接在@RabbitListener上配置

   @Component
   public class SpringBootMsqConsumer {
       @RabbitListener(queues = "spring-boot-direct-queue",concurrency = "5-10")
       public void receive(Message message) {
           System.out.println("receive message:" + new String(message.getBody()));
       }
   }
   

   //@RabbitListener(queues = {"kinson"}, concurrency =  "2")
   @RabbitListener(queues = {"kinson"}, concurency="min-max") // 最小并发数是5，最大并发是10
   public void receiver(Message msg, Channel channel) throws InterruptedException {
   
   //        Thread.sleep(10000);
       byte[] messageBytes = msg.getBody();
   
       if (messageBytes != null && messageBytes.length > 0) {
           //打印数据
           String message = new String(msg.getBody(), StandardCharsets.UTF_8);
           log.info("【消3】:{}", message);
       }
   }
   

   @RabbitListener(
            bindings = @QueueBinding(
                    value = @Queue(value = "myQueue1"),
                    exchange = @Exchange(value = "myExchange1"),
                    key = "routingKey1"
            ),
            concurrency =  "10"
    )
    public void process1(Message message) throws Exception {
        // 执行程序
    }
   

   利用@RabbitListener中的concurrency属性进行指定就行，例如上面的concurrency = “5-10”

   就表示最小5个，最大10个consumer。启动消费端应用后，找到spring-boot-direct-queue这个队列的consumer，会发现有5个。
   
   这5个消费者都可以从spring-boot-direct-queue这个队列中获取消息，加快队列中消息的消费速度，提高吞吐量。

限流

我们经过压测，来判断consumer的消费能力，如果单位时间内，consumer到达的消息太多，也可能把消费者压垮。
得到压测数据后，可以在@RabbitListener中配置prefetch count。

import org.springframework.amqp.core.Message;
import org.springframework.amqp.rabbit.annotation.RabbitListener;
import org.springframework.stereotype.Component;

@Component
public class SpringBootMsqConsumer {
    @RabbitListener(queues = "spring-boot-direct-queue",concurrency = "5-10",containerFactory = "mqConsumerlistenerContainer")
    public void receive(Message message) {
        System.out.println("receive message:" + new String(message.getBody()));
    }
}

import org.springframework.amqp.rabbit.config.SimpleRabbitListenerContainerFactory;
import org.springframework.amqp.rabbit.connection.CachingConnectionFactory;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;

@Configuration
public class RabbitMqConfig {

    @Autowired
    private CachingConnectionFactory connectionFactory;

    @Bean(name = "mqConsumerlistenerContainer")
    public SimpleRabbitListenerContainerFactory mqConsumerlistenerContainer(){
        SimpleRabbitListenerContainerFactory factory = new SimpleRabbitListenerContainerFactory();
        factory.setConnectionFactory(connectionFactory);
        factory.setPrefetchCount(50);
        return factory;
    }
}

factory.setPrefetchCount(50);，就是用于设置prefetch count的，启动后，会在spring-boot-direct-queue队列的consumer中体现出来。

配置成功后，consumer单位时间内接收到消息就是50条。

---

@Autowired
CachingConnectionFactory cachingConnectionFactory;

@Bean(name="limitContainerFactory")
public SimpleRabbitListenerContainerFactory simpleRabbitListenerContainerFactory(){
    SimpleRabbitListenerContainerFactory factory=new SimpleRabbitListenerContainerFactory();
    factory.setConnectionFactory(cachingConnectionFactory);
    factory.setConcurrentConsumers(60); // 并发消费者数量
    factory.setPrefetchCount(3);  // 每次最多拿3个，等这三个处理完之后，再去队列中拿第二批，起到了限流的作用
    return factory;
}

---

@Bean("limitContainerFactory")
public SimpleRabbitListenerContainerFactory pointTaskContainerFactory(SimpleRabbitListenerContainerFactoryConfigurer configurer, ConnectionFactory connectionFactory) {
    SimpleRabbitListenerContainerFactory factory = new SimpleRabbitListenerContainerFactory();
    factory.setPrefetchCount(2); //每个消费者预取数量
    factory.setConcurrentConsumers(60); //并发消费者数量
    configurer.configure(factory, connectionFactory);
    return factory;
}

---

@RabbitHandler
@RabbitListener(queues = QueueContent.MESSAGE_QUEUE_NAME,containerFactory = "limitContainerFactory")
public void handler(String msg,Channel channel, Message message) throws IOException {
	...... 
}