一、RabbitMQ介绍

1.1 现存问题
  • 服务调用:两个服务调用时,我们可以通过传统的HTTP方式,让服务A直接去调用服务B的接口,但是这种方式是同步的方式,虽然可以采用SpringBoot提供的@Async注解实现异步调用,但是这种方式无法确保请求一定回访问到服务B的接口。那如何保证服务A的请求信息一定能送达到服务B去完成一些业务操作呢?| 如何实现异步调用
  • 海量请求:在我们在做一些秒杀业务时,可能会在某个时间点突然出现大量的并发请求,这可能已经远远超过服务器的并发瓶颈,这时我们需要做一些削峰的操作,也就是将大量的请求缓冲到一个队列中,然后慢慢的消费掉。如何提供一个可以存储千万级别请求的队列呢?
  • 在微服务架构下,可能一个业务会出现同时调用多个其他服务的场景,而且这些服务之间一般会用到Feign的方式进行轻量级的通讯,如果存在一个业务,用户创建订单成功后,还需要去给用户添加积分、通知商家、通知物流系统、扣减商品库存,而在执行这个操作时,如果任意一个服务出现了问题,都会导致整体的下单业务失败,并且会导致给用户反馈的时间延长。这时就造成了服务之间存在一个较高的耦合性的问题。如何可以降低服务之间的耦合性呢?
1.2 处理问题

RabbitMQ就可以解决上述的全部问题

  • 服务之间如何想实现可靠的异步调用,可以通过RabbitMQ的方式实现,服务A只需要保证可以把消息发送到RabbitMQ的队列中,服务B就一定会消费到队列中的消息只不过会存在一定的延时。| 异步访问
  • 忽然的海量请求可以存储在RabbitMQ的队列中,然后由消费者慢慢消费掉,RabbitMQ的队列本身就可以存储上千万条消息
  • 在调用其他服务时,如果允许延迟效果的出现,可以将消息发送到RabbitMQ中,再由消费者慢慢消费| 服务解耦
1.3 RabbitMQ介绍

百度百科:

RabbitMQ是实现了高级消息队列协议(AMQP)的开源消息代理软件(亦称面向消息的中间件)。RabbitMQ服务器是用Erlang语言编写的,而集群和故障转移是构建在开放电信平台框架上的。所有主要的编程语言均有与代理接口通讯的客户端库。

首先RabbitMQ基于AMQP协议开发,所以很多基于AMQP协议的功能RabbitMQ都是支持的,比如SpringCloud中的消息总线bus

其次RabbitMQ是基于Erlang编写,这是也是RabbitMQ天生的优势,Erlang被称为面向并发编程的语言,并发能力极强,在众多的MQ中,RabbitMQ的延迟特别低,在微秒级别,所以一般的业务处理RabbitMQ比Kafka和RocketMQ更有优势。

最后RabbitMQ提供自带了图形化界面,操作方便,还自带了多种集群模式,可以保证RabbitMQ的高可用,并且SpringBoot默认就整合RabbitMQ,使用简单方便。

二、RabbitMQ安装


2.1 安装RabbitMQ

这里推荐搭建采用Docker的方式在Linux中安装RabbitMQ,如果对Docker不了解,推荐去学习一下Docker的应用,不然学习其他的知识时,安装的成本都特别高,这里我们就采用Docker的方式安装RabbitMQ。

直接使用docker-compose.yml文件即可安装RabbitMQ服务

version: '3.1'services:rabbitmq:restart: alwaysimage: daocloud.io/library/rabbitmq:3.8.8volumes:- ./data/:/var/lib/rabbitmq/- ./log/:/var/log/rabbitmq/log/ports:- 15672:15672- 5672:5672

执行 docker-compose up -d运行

测试效果:curl localhost:5672

查看效果
2.2 开启图形化界面

默认情况下,当前镜像的图形化界面默认没有开启,需要进入到容器内部开启图形化管理界面

启动图形化界面插件

通过浏览器访问15672,查看图形化界面

查看登录页面

默认用户和密码均为:guest,查看首页

查看首页

三、RabbitMQ构架

RabbitMQ的架构可以查看官方地址:https://rabbitmq.com/tutorials/amqp-concepts.html

官方简单架构

可以看出RabbitMQ中主要分为三个角色:

  • Publisher:消息的发布者,将消息发布到RabbitMQ中的Exchange
  • RabbitMQ服务:Exchange接收Publisher的消息,并且根据Routes策略将消息转发到Queue中
  • Consumer:消息的消费者,监听Queue中的消息并进行消费

官方提供的架构图相对简洁,我们可以自己画一份相对完整一些的架构图:

RabbitMQ架构图

可以看出Publisher和Consumer都是单独和RabbitMQ服务中某一个Virtual Host建立Connection的客户端

后续通过Connection可以构建Channel通道,用来发布、接收消息

一个Virtual Host中可以有多个Exchange和Queue,Exchange可以同时绑定多个Queue

在基于架构图查看图形化界面,会更加清晰

图形化界面信息

四、RabbitMQ通讯方式

RabbitMQ提供了很多中通讯方式,依然可以去官方查看:https://rabbitmq.com/getstarted.html

七种通讯方式
4.1 RabbitMQ提供的通讯方式
  • Hello World!:为了入门操作!
  • Work queues:一个队列被多个消费者消费
  • Publish/Subscribe:手动创建Exchange(FANOUT)
  • Routing:手动创建Exchange(DIRECT)
  • Topics:手动创建Exchange(TOPIC)
  • RPC:RPC方式
  • Publisher Confirms:保证消息可靠性
4.2 构建Connection工具类
  • 导入依赖:amqp-client,junit

    <dependencies><dependency><groupId>com.rabbitmq</groupId><artifactId>amqp-client</artifactId><version>5.9.0</version></dependency><dependency><groupId>junit</groupId><artifactId>junit</artifactId><version>4.12</version></dependency></dependencies>
  • 构建工具类:

    package com.mashibing.util;import com.rabbitmq.client.Connection;import com.rabbitmq.client.ConnectionFactory;import java.io.IOException;import java.util.concurrent.TimeoutException;/** * @author zjw * @description */public class RabbitMQConnectionUtil {public static final String RABBITMQ_HOST = "192.168.11.32";public static final int RABBITMQ_PORT = 5672;public static final String RABBITMQ_USERNAME = "guest";public static final String RABBITMQ_PASSWORD = "guest";public static final String RABBITMQ_VIRTUAL_HOST = "/";/** * 构建RabbitMQ的连接对象 * @return */public static Connection getConnection() throws Exception {//1. 创建Connection工厂ConnectionFactory factory = new ConnectionFactory();//2. 设置RabbitMQ的连接信息factory.setHost(RABBITMQ_HOST);factory.setPort(RABBITMQ_PORT);factory.setUsername(RABBITMQ_USERNAME);factory.setPassword(RABBITMQ_PASSWORD);factory.setVirtualHost(RABBITMQ_VIRTUAL_HOST);//3. 返回连接对象Connection connection = factory.newConnection();return connection;}}
4.3 Hello World
通讯方式

生产者:

package com.mashibing.helloworld;import com.mashibing.util.RabbitMQConnectionUtil;import com.rabbitmq.client.Channel;import com.rabbitmq.client.Connection;import org.junit.Test;/** * @author zjw * @description * @date 2022/1/24 22:54 */public class Publisher {public static final String QUEUE_NAME = "hello";@Testpublic void publish() throws Exception {//1. 获取连接对象Connection connection = RabbitMQConnectionUtil.getConnection();//2. 构建ChannelChannel channel = connection.createChannel();//3. 构建队列channel.queueDeclare(QUEUE_NAME,false,false,false,null);//4. 发布消息String message = "Hello World!";channel.basicPublish("",QUEUE_NAME,null,message.getBytes());System.out.println("消息发送成功!");}}

消费者:

package com.mashibing.helloworld;import com.mashibing.util.RabbitMQConnectionUtil;import com.rabbitmq.client.*;import org.junit.Test;import java.io.IOException;/** * @author zjw * @description * @date 2022/1/24 23:02 */public class Consumer {@Testpublic void consume() throws Exception {//1. 获取连接对象Connection connection = RabbitMQConnectionUtil.getConnection();//2. 构建ChannelChannel channel = connection.createChannel();//3. 构建队列channel.queueDeclare(Publisher.QUEUE_NAME,false,false,false,null);//4. 监听消息DefaultConsumer callback = new DefaultConsumer(channel){@Overridepublic void handleDelivery(String consumerTag, Envelope envelope, AMQP.BasicProperties properties, byte[] body) throws IOException {System.out.println("消费者获取到消息:" + new String(body,"UTF-8"));}};channel.basicConsume(Publisher.QUEUE_NAME,true,callback);System.out.println("开始监听队列");System.in.read();}}
4.4 Work Queues
WorkQueues需要学习的内容
  • 生产者:生产者和Hello World的形式是一样的,都是将消息推送到默认交换机。

  • 消费者:让消费者关闭自动ack,并且设置消息的流控,最终实现消费者可以尽可能去多消费消息

    package com.mashibing.workqueues;import com.mashibing.util.RabbitMQConnectionUtil;import com.rabbitmq.client.*;import org.junit.Test;import java.io.IOException;/** * @author zjw * @description * @date 2022/1/25 19:52 */public class Consumer {@Testpublic void consume1() throws Exception {//1. 获取连接对象Connection connection = RabbitMQConnectionUtil.getConnection();//2. 构建ChannelChannel channel = connection.createChannel();//3. 构建队列channel.queueDeclare(Publisher.QUEUE_NAME,false,false,false,null);//3.5 设置消息的流控channel.basicQos(3);//4. 监听消息DefaultConsumer callback = new DefaultConsumer(channel){@Overridepublic void handleDelivery(String consumerTag, Envelope envelope, AMQP.BasicProperties properties, byte[] body) throws IOException {try {Thread.sleep(100);} catch (InterruptedException e) {e.printStackTrace();}System.out.println("消费者1号-获取到消息:" + new String(body,"UTF-8"));channel.basicAck(envelope.getDeliveryTag(),false);}};channel.basicConsume(Publisher.QUEUE_NAME,false,callback);System.out.println("开始监听队列");System.in.read();}@Testpublic void consume2() throws Exception {//1. 获取连接对象Connection connection = RabbitMQConnectionUtil.getConnection();//2. 构建ChannelChannel channel = connection.createChannel();//3. 构建队列channel.queueDeclare(Publisher.QUEUE_NAME,false,false,false,null);channel.basicQos(3);//4. 监听消息DefaultConsumer callback = new DefaultConsumer(channel){@Overridepublic void handleDelivery(String consumerTag, Envelope envelope, AMQP.BasicProperties properties, byte[] body) throws IOException {try {Thread.sleep(1000);} catch (InterruptedException e) {e.printStackTrace();}System.out.println("消费者2号-获取到消息:" + new String(body,"UTF-8"));channel.basicAck(envelope.getDeliveryTag(),false);}};channel.basicConsume(Publisher.QUEUE_NAME,false,callback);System.out.println("开始监听队列");System.in.read();}}
4.5 Publish/Subscribe
自定义一个交换机

生产者:自行构建Exchange并绑定指定队列(FANOUT类型)

package com.mashibing.pubsub;import com.mashibing.util.RabbitMQConnectionUtil;import com.rabbitmq.client.BuiltinExchangeType;import com.rabbitmq.client.Channel;import com.rabbitmq.client.Connection;import org.junit.Test;/** * @author zjw * @description * @date 2022/1/25 20:08 */public class Publisher {public static final String EXCHANGE_NAME = "pubsub";public static final String QUEUE_NAME1 = "pubsub-one";public static final String QUEUE_NAME2 = "pubsub-two";@Testpublic void publish() throws Exception {//1. 获取连接对象Connection connection = RabbitMQConnectionUtil.getConnection();//2. 构建ChannelChannel channel = connection.createChannel();//3. 构建交换机channel.exchangeDeclare(EXCHANGE_NAME, BuiltinExchangeType.FANOUT);//4. 构建队列channel.queueDeclare(QUEUE_NAME1,false,false,false,null);channel.queueDeclare(QUEUE_NAME2,false,false,false,null);//5. 绑定交换机和队列,使用的是FANOUT类型的交换机,绑定方式是直接绑定channel.queueBind(QUEUE_NAME1,EXCHANGE_NAME,"");channel.queueBind(QUEUE_NAME2,EXCHANGE_NAME,"");//6. 发消息到交换机channel.basicPublish(EXCHANGE_NAME,"45jk6h645jk",null,"publish/subscribe!".getBytes());System.out.println("消息成功发送!");}}
4.6 Routing
DIRECT类型Exchange

生产者:在绑定Exchange和Queue时,需要指定好routingKey,同时在发送消息时,也指定routingKey,只有routingKey一致时,才会把指定的消息路由到指定的Queue

package com.mashibing.routing;import com.mashibing.util.RabbitMQConnectionUtil;import com.rabbitmq.client.BuiltinExchangeType;import com.rabbitmq.client.Channel;import com.rabbitmq.client.Connection;import org.junit.Test;/** * @author zjw * @description * @date 2022/1/25 20:20 */public class Publisher {public static final String EXCHANGE_NAME = "routing";public static final String QUEUE_NAME1 = "routing-one";public static final String QUEUE_NAME2 = "routing-two";@Testpublic void publish() throws Exception {//1. 获取连接对象Connection connection = RabbitMQConnectionUtil.getConnection();//2. 构建ChannelChannel channel = connection.createChannel();//3. 构建交换机channel.exchangeDeclare(EXCHANGE_NAME, BuiltinExchangeType.DIRECT);//4. 构建队列channel.queueDeclare(QUEUE_NAME1,false,false,false,null);channel.queueDeclare(QUEUE_NAME2,false,false,false,null);//5. 绑定交换机和队列channel.queueBind(QUEUE_NAME1,EXCHANGE_NAME,"ORANGE");channel.queueBind(QUEUE_NAME2,EXCHANGE_NAME,"BLACK");channel.queueBind(QUEUE_NAME2,EXCHANGE_NAME,"GREEN");//6. 发消息到交换机channel.basicPublish(EXCHANGE_NAME,"ORANGE",null,"大橙子!".getBytes());channel.basicPublish(EXCHANGE_NAME,"BLACK",null,"黑布林大狸子".getBytes());channel.basicPublish(EXCHANGE_NAME,"WHITE",null,"小白兔!".getBytes());System.out.println("消息成功发送!");}}
4.7 Topic
Topic模式

生产者:TOPIC类型可以编写带有特殊意义的routingKey的绑定方式

package com.mashibing.topics;import com.mashibing.util.RabbitMQConnectionUtil;import com.rabbitmq.client.BuiltinExchangeType;import com.rabbitmq.client.Channel;import com.rabbitmq.client.Connection;import org.junit.Test;/** * @author zjw * @description * @date 2022/1/25 20:28 */public class Publisher {public static final String EXCHANGE_NAME = "topic";public static final String QUEUE_NAME1 = "topic-one";public static final String QUEUE_NAME2 = "topic-two";@Testpublic void publish() throws Exception {//1. 获取连接对象Connection connection = RabbitMQConnectionUtil.getConnection();//2. 构建ChannelChannel channel = connection.createChannel();//3. 构建交换机channel.exchangeDeclare(EXCHANGE_NAME, BuiltinExchangeType.TOPIC);//4. 构建队列channel.queueDeclare(QUEUE_NAME1,false,false,false,null);channel.queueDeclare(QUEUE_NAME2,false,false,false,null);//5. 绑定交换机和队列,// TOPIC类型的交换机在和队列绑定时,需要以aaa.bbb.ccc..方式编写routingkey// 其中有两个特殊字符:*(相当于占位符),#(相当通配符)channel.queueBind(QUEUE_NAME1,EXCHANGE_NAME,"*.orange.*");channel.queueBind(QUEUE_NAME2,EXCHANGE_NAME,"*.*.rabbit");channel.queueBind(QUEUE_NAME2,EXCHANGE_NAME,"lazy.#");//6. 发消息到交换机channel.basicPublish(EXCHANGE_NAME,"big.orange.rabbit",null,"大橙兔子!".getBytes());channel.basicPublish(EXCHANGE_NAME,"small.white.rabbit",null,"小白兔".getBytes());channel.basicPublish(EXCHANGE_NAME,"lazy.dog.dog.dog.dog.dog.dog",null,"懒狗狗狗狗狗狗".getBytes());System.out.println("消息成功发送!");}}
4.8 RPC(了解)

因为两个服务在交互时,可以尽量做到Client和Server的解耦,通过RabbitMQ进行解耦操作

需要让Client发送消息时,携带两个属性:

  • replyTo告知Server将相应信息放到哪个队列
  • correlationId告知Server发送相应消息时,需要携带位置标示来告知Client响应的信息
RPC方式

客户端:

package com.mashibing.rpc;import com.mashibing.util.RabbitMQConnectionUtil;import com.rabbitmq.client.*;import org.junit.Test;import java.io.IOException;import java.util.UUID;/** * @author zjw * @description * @date 2022/2/8 20:03 */public class Publisher {public static final String QUEUE_PUBLISHER = "rpc_publisher";public static final String QUEUE_CONSUMER = "rpc_consumer";@Testpublic void publish() throws Exception {//1. 获取连接对象Connection connection = RabbitMQConnectionUtil.getConnection();//2. 构建ChannelChannel channel = connection.createChannel();//3. 构建队列channel.queueDeclare(QUEUE_PUBLISHER,false,false,false,null);channel.queueDeclare(QUEUE_CONSUMER,false,false,false,null);//4. 发布消息String message = "Hello RPC!";String uuid = UUID.randomUUID().toString();AMQP.BasicProperties props = new AMQP.BasicProperties().builder().replyTo(QUEUE_CONSUMER).correlationId(uuid).build();channel.basicPublish("",QUEUE_PUBLISHER,props,message.getBytes());channel.basicConsume(QUEUE_CONSUMER,false,new DefaultConsumer(channel){@Overridepublic void handleDelivery(String consumerTag, Envelope envelope, AMQP.BasicProperties properties, byte[] body) throws IOException {String id = properties.getCorrelationId();if(id != null && id.equalsIgnoreCase(uuid)){System.out.println("接收到服务端的响应:" + new String(body,"UTF-8"));}channel.basicAck(envelope.getDeliveryTag(),false);}});System.out.println("消息发送成功!");System.in.read();}}

服务端:

package com.mashibing.rpc;import com.mashibing.helloworld.Publisher;import com.mashibing.util.RabbitMQConnectionUtil;import com.rabbitmq.client.*;import org.junit.Test;import java.io.IOException;/** * @author zjw * @description * @date 2022/1/24 23:02 */public class Consumer {public static final String QUEUE_PUBLISHER = "rpc_publisher";public static final String QUEUE_CONSUMER = "rpc_consumer";@Testpublic void consume() throws Exception {//1. 获取连接对象Connection connection = RabbitMQConnectionUtil.getConnection();//2. 构建ChannelChannel channel = connection.createChannel();//3. 构建队列channel.queueDeclare(QUEUE_PUBLISHER,false,false,false,null);channel.queueDeclare(QUEUE_CONSUMER,false,false,false,null);//4. 监听消息DefaultConsumer callback = new DefaultConsumer(channel){@Overridepublic void handleDelivery(String consumerTag, Envelope envelope, AMQP.BasicProperties properties, byte[] body) throws IOException {System.out.println("消费者获取到消息:" + new String(body,"UTF-8"));String resp = "获取到了client发出的请求,这里是响应的信息";String respQueueName = properties.getReplyTo();String uuid = properties.getCorrelationId();AMQP.BasicProperties props = new AMQP.BasicProperties().builder().correlationId(uuid).build();channel.basicPublish("",respQueueName,props,resp.getBytes());channel.basicAck(envelope.getDeliveryTag(),false);}};channel.basicConsume(QUEUE_PUBLISHER,false,callback);System.out.println("开始监听队列");System.in.read();}}

五、SpringBoot操作RabbitMQ

5.1 SpringBoot声明信息
  • 创建项目

  • 导入依赖

    <dependency><groupId>org.springframework.boot</groupId><artifactId>spring-boot-starter-amqp</artifactId></dependency>
  • 配置RabbitMQ信息

    spring:rabbitmq:host: 192.168.11.32port: 5672username: guestpassword: guestvirtual-host: /
  • 声明交换机&队列

    package com.mashibing.rabbitmqboot.config;import org.springframework.amqp.core.*;import org.springframework.context.annotation.Bean;import org.springframework.context.annotation.Configuration;/** * @author zjw * @description * @date 2022/2/8 20:25 */@Configurationpublic class RabbitMQConfig {public static final String EXCHANGE = "boot-exchange";public static final String QUEUE = "boot-queue";public static final String ROUTING_KEY = "*.black.*";@Beanpublic Exchange bootExchange(){// channel.DeclareExchangereturn ExchangeBuilder.topicExchange(EXCHANGE).build();}@Beanpublic Queue bootQueue(){return QueueBuilder.durable(QUEUE).build();}@Beanpublic Binding bootBinding(Exchange bootExchange,Queue bootQueue){return BindingBuilder.bind(bootQueue).to(bootExchange).with(ROUTING_KEY).noargs();}}
5.2 生产者操作
package com.mashibing.rabbitmqboot;import com.mashibing.rabbitmqboot.config.RabbitMQConfig;import org.junit.jupiter.api.Test;import org.springframework.amqp.AmqpException;import org.springframework.amqp.core.Message;import org.springframework.amqp.core.MessagePostProcessor;import org.springframework.amqp.rabbit.core.RabbitTemplate;import org.springframework.beans.factory.annotation.Autowired;import org.springframework.boot.test.context.SpringBootTest;/** * @author zjw * @description * @date 2022/2/8 21:05 */@SpringBootTestpublic class PublisherTest {@Autowiredpublic RabbitTemplate rabbitTemplate;@Testpublic void publish(){rabbitTemplate.convertAndSend(RabbitMQConfig.EXCHANGE,"big.black.dog","message");System.out.println("消息发送成功");}@Testpublic void publishWithProps(){rabbitTemplate.convertAndSend(RabbitMQConfig.EXCHANGE, "big.black.dog", "messageWithProps", new MessagePostProcessor() {@Overridepublic Message postProcessMessage(Message message) throws AmqpException {message.getMessageProperties().setCorrelationId("123");return message;}});System.out.println("消息发送成功");}} #### 5.3 消费者操作```javapackage com.mashibing.rabbitmqboot;import com.mashibing.rabbitmqboot.config.RabbitMQConfig;import com.rabbitmq.client.Channel;import org.junit.jupiter.api.Test;import org.springframework.amqp.core.Message;import org.springframework.amqp.rabbit.annotation.RabbitListener;import org.springframework.boot.test.context.SpringBootTest;import org.springframework.context.annotation.Configuration;import org.springframework.stereotype.Component;import java.io.IOException;/** * @author zjw * @description * @date 2022/2/8 21:11 */@Componentpublic class ConsumeListener {@RabbitListener(queues = RabbitMQConfig.QUEUE)public void consume(String msg, Channel channel, Message message) throws IOException {System.out.println("队列的消息为:" + msg);String correlationId = message.getMessageProperties().getCorrelationId();System.out.println("唯一标识为:" + correlationId);channel.basicAck(message.getMessageProperties().getDeliveryTag(),false);}}

六、RabbitMQ保证消息可靠性

6.1 保证消息一定送达到Exchange

Confirm机制

可以通过Confirm效果保证消息一定送达到Exchange,官方提供了三种方式,选择了对于效率影响最低的异步回调的效果

//4. 开启confirmschannel.confirmSelect();//5. 设置confirms的异步回调channel.addConfirmListener(new ConfirmListener() {@Overridepublic void handleAck(long deliveryTag, boolean multiple) throws IOException {System.out.println("消息成功的发送到Exchange!");}@Overridepublic void handleNack(long deliveryTag, boolean multiple) throws IOException {System.out.println("消息没有发送到Exchange,尝试重试,或者保存到数据库做其他补偿操作!");}});
6.2 保证消息可以路由到Queue

Return机制

为了保证Exchange上的消息一定可以送达到Queue

//6. 设置Return回调,确认消息是否路由到了Queuechannel.addReturnListener(new ReturnListener() {@Overridepublic void handleReturn(int replyCode, String replyText, String exchange, String routingKey, AMQP.BasicProperties properties, byte[] body) throws IOException {System.out.println("消息没有路由到指定队列,做其他的补偿措施!!");}});//7. 在发送消息时,将basicPublish方法参数中的mandatory设置为true,即可开启Return机制,当消息没有路由到队列中时,就会执行return回调
6.3 保证Queue可以持久化消息

DeliveryMode设置消息持久化

DeliveryMode设置为2代表持久化,如果设置为1,就代表不会持久化。

//7. 设置消息持久化AMQP.BasicProperties props = new AMQP.BasicProperties().builder().deliveryMode(2).build();//7. 发布消息channel.basicPublish("","confirms",true,props,message.getBytes());
6.4 保证消费者可以正常消费消息

详情看WorkQueue模式

6.5 SpringBoot实现上述操作
6.5.1 Confirm
  • 编写配置文件开启Confirm机制

    spring:rabbitmq:publisher-confirm-type: correlated# 新版本publisher-confirms: true# 老版本 
  • 在发送消息时,配置RabbitTemplate

    @Testpublic void publishWithConfirms() throws IOException {rabbitTemplate.setConfirmCallback(new RabbitTemplate.ConfirmCallback() {@Overridepublic void confirm(CorrelationData correlationData, boolean ack, String cause) {if(ack){System.out.println("消息已经送达到交换机!!");}else{System.out.println("消息没有送达到Exchange,需要做一些补偿操作!!retry!!!");}}});rabbitTemplate.convertAndSend(RabbitMQConfig.EXCHANGE,"big.black.dog","message");System.out.println("消息发送成功");System.in.read();}
6.5.2 Return
  • 编写配置文件开启Return机制

    spring:rabbitmq:publisher-returns: true # 开启Return机制
  • 在发送消息时,配置RabbitTemplate

    @Testpublic void publishWithReturn() throws IOException {// 新版本用 setReturnsCallback ,老版本用setReturnCallbackrabbitTemplate.setReturnsCallback(new RabbitTemplate.ReturnsCallback() {@Overridepublic void returnedMessage(ReturnedMessage returned) {String msg = new String(returned.getMessage().getBody());System.out.println("消息:" + msg + "路由队列失败!!做补救操作!!");}});rabbitTemplate.convertAndSend(RabbitMQConfig.EXCHANGE,"big.black.dog","message");System.out.println("消息发送成功");System.in.read();}
6.5.3 消息持久化
@Testpublic void publishWithBasicProperties() throws IOException {rabbitTemplate.convertAndSend(RabbitMQConfig.EXCHANGE, "big.black.dog", "message", new MessagePostProcessor() {@Overridepublic Message postProcessMessage(Message message) throws AmqpException {// 设置消息的持久化!message.getMessageProperties().setDeliveryMode(MessageDeliveryMode.PERSISTENT);return message;}});System.out.println("消息发送成功");}

七、RabbitMQ死信队列&延迟交换机

7.1 什么是死信
死信&死信队列

死信队列的应用:

  • 基于死信队列在队列消息已满的情况下,消息也不会丢失
  • 实现延迟消费的效果。比如:下订单时,有15分钟的付款时间
7.2 实现死信队列
7.2.1 准备Exchange&Queue
 package com.mashibing.rabbitmqboot.config;import org.springframework.amqp.core.*;import org.springframework.context.annotation.Bean;import org.springframework.context.annotation.Configuration;/** * @author zjw * @description * @date 2022/2/10 15:04 */@Configurationpublic class DeadLetterConfig {public static final String NORMAL_EXCHANGE = "normal-exchange";public static final String NORMAL_QUEUE = "normal-queue";public static final String NORMAL_ROUTING_KEY = "normal.#";public static final String DEAD_EXCHANGE = "dead-exchange";public static final String DEAD_QUEUE = "dead-queue";public static final String DEAD_ROUTING_KEY = "dead.#";@Beanpublic Exchange normalExchange(){return ExchangeBuilder.topicExchange(NORMAL_EXCHANGE).build();}@Beanpublic Queue normalQueue(){return QueueBuilder.durable(NORMAL_QUEUE).deadLetterExchange(DEAD_EXCHANGE).deadLetterRoutingKey("dead.abc").build();}@Beanpublic Binding normalBinding(Queue normalQueue,Exchange normalExchange){return BindingBuilder.bind(normalQueue).to(normalExchange).with(NORMAL_ROUTING_KEY).noargs();}@Beanpublic Exchange deadExchange(){return ExchangeBuilder.topicExchange(DEAD_EXCHANGE).build();}@Beanpublic Queue deadQueue(){return QueueBuilder.durable(DEAD_QUEUE).build();}@Beanpublic Binding deadBinding(Queue deadQueue,Exchange deadExchange){return BindingBuilder.bind(deadQueue).to(deadExchange).with(DEAD_ROUTING_KEY).noargs();}}
7.2.2 实现效果
  • 基于消费者进行reject或者nack实现死信效果

    package com.mashibing.rabbitmqboot;import com.mashibing.rabbitmqboot.config.DeadLetterConfig;import com.rabbitmq.client.Channel;import org.springframework.amqp.core.Message;import org.springframework.amqp.rabbit.annotation.RabbitListener;import org.springframework.stereotype.Component;import java.io.IOException;/** * @author zjw * @description * @date 2022/2/10 15:17 */@Componentpublic class DeadListener {@RabbitListener(queues = DeadLetterConfig.NORMAL_QUEUE)public void consume(String msg, Channel channel, Message message) throws IOException {System.out.println("接收到normal队列的消息:" + msg);channel.basicReject(message.getMessageProperties().getDeliveryTag(),false);channel.basicNack(message.getMessageProperties().getDeliveryTag(),false,false);}}
  • 消息的生存时间

    • 给消息设置生存时间

      @Testpublic void publishExpire(){String msg = "dead letter expire";rabbitTemplate.convertAndSend(DeadLetterConfig.NORMAL_EXCHANGE, "normal.abc", msg, new MessagePostProcessor() {@Overridepublic Message postProcessMessage(Message message) throws AmqpException {message.getMessageProperties().setExpiration("5000");return message;}});}
    • 给队列设置消息的生存时间

      @Beanpublic Queue normalQueue(){return QueueBuilder.durable(NORMAL_QUEUE).deadLetterExchange(DEAD_EXCHANGE).deadLetterRoutingKey("dead.abc").ttl(10000).build();}
  • 设置Queue中的消息最大长度

    @Beanpublic Queue normalQueue(){return QueueBuilder.durable(NORMAL_QUEUE).deadLetterExchange(DEAD_EXCHANGE).deadLetterRoutingKey("dead.abc").maxLength(1).build();}

    只要Queue中已经有一个消息,如果再次发送一个消息,这个消息会变为死信!

7.3 延迟交换机

下载地址:https://github.com/rabbitmq/rabbitmq-delayed-message-exchange/releases/tag/3.8.9

死信队列实现延迟消费时,如果延迟时间比较复杂,比较多,直接使用死信队列时,需要创建大量的队列还对应不同的时间,可以采用延迟交换机来解决这个问题。

  • 构建延迟交换机

    package com.mashibing.rabbitmqboot.config;import org.springframework.amqp.core.*;import org.springframework.context.annotation.Bean;import org.springframework.context.annotation.Configuration;import java.util.HashMap;import java.util.Map;/** * @author zjw * @description */@Configurationpublic class DelayedConfig {public static final String DELAYED_EXCHANGE = "delayed-exchange";public static final String DELAYED_QUEUE = "delayed-queue";public static final String DELAYED_ROUTING_KEY = "delayed.#";@Beanpublic Exchange delayedExchange(){Map<String, Object> arguments = new HashMap<>();arguments.put("x-delayed-type","topic");Exchange exchange = new CustomExchange(DELAYED_EXCHANGE,"x-delayed-message",true,false,arguments);return exchange;}@Beanpublic Queue delayedQueue(){return QueueBuilder.durable(DELAYED_QUEUE).build();}@Beanpublic Binding delayedBinding(Queue delayedQueue,Exchange delayedExchange){return BindingBuilder.bind(delayedQueue).to(delayedExchange).with(DELAYED_ROUTING_KEY).noargs();}}
  • 发送消息

    package com.mashibing.rabbitmqboot;import com.mashibing.rabbitmqboot.config.DelayedConfig;import org.junit.jupiter.api.Test;import org.springframework.amqp.AmqpException;import org.springframework.amqp.core.Message;import org.springframework.amqp.core.MessagePostProcessor;import org.springframework.amqp.rabbit.core.RabbitTemplate;import org.springframework.beans.factory.annotation.Autowired;import org.springframework.boot.test.context.SpringBootTest;/** * @author zjw * @description */@SpringBootTestpublic class DelayedPublisherTest {@Autowiredprivate RabbitTemplate rabbitTemplate;@Testpublic void publish(){rabbitTemplate.convertAndSend(DelayedConfig.DELAYED_EXCHANGE, "delayed.abc", "xxxx", new MessagePostProcessor() {@Overridepublic Message postProcessMessage(Message message) throws AmqpException {message.getMessageProperties().setDelay(30000);return message;}});}}

八、RabbitMQ的集群

RabbitMQ的镜像模式

RabbitMQ的集群

高可用

提升RabbitMQ的效率

搭建RabbitMQ集群

  • 准备两台虚拟机(克隆)

  • 准备RabbitMQ的yml文件

    rabbitmq1:

    version: '3.1'services:rabbitmq1:image: rabbitmq:3.8.5-management-alpinecontainer_name: rabbitmq1hostname: rabbitmq1extra_hosts:- "rabbitmq1:192.168.11.32"- "rabbitmq2:192.168.11.33"environment: - RABBITMQ_ERLANG_COOKIE=SDJHFGDFFSports:- 5672:5672- 15672:15672- 4369:4369- 25672:25672

    rabbitmq2:

    version: '3.1'services:rabbitmq2:image: rabbitmq:3.8.5-management-alpinecontainer_name: rabbitmq2hostname: rabbitmq2extra_hosts:- "rabbitmq1:192.168.11.32"- "rabbitmq2:192.168.11.33"environment: - RABBITMQ_ERLANG_COOKIE=SDJHFGDFFSports:- 5672:5672- 15672:15672- 4369:4369- 25672:25672

    准备完毕之后,启动两台RabbitMQ

    启动效果
  • 让RabbitMQ服务实现join操作

    需要四个命令完成join操作

    让rabbitmq2 join rabbitmq1,需要进入到rabbitmq2的容器内部,去执行下述命令

    rabbitmqctl stop_apprabbitmqctl reset rabbitmqctl join_cluster rabbit@rabbitmq1rabbitmqctl start_app

    执行成功后:

    执行成功后
  • 设置镜像模式

    在指定的RabbitMQ服务中设置好镜像策略即可

    镜像模式

九、RabbitMQ其他内容

9.1 Headers类型Exchange

headers就是一个基于key-value的方式,让Exchange和Queue绑定的到一起的一种规则

相比Topic形式,可以采用的类型更丰富。

headers绑定方式

具体实现方式

package com.mashibing.headers;import com.mashibing.util.RabbitMQConnectionUtil;import com.rabbitmq.client.AMQP;import com.rabbitmq.client.BuiltinExchangeType;import com.rabbitmq.client.Channel;import com.rabbitmq.client.Connection;import org.junit.Test;import java.util.HashMap;import java.util.Map;/** * @author zjw * @description */public class Publisher {public static final String HEADER_EXCHANGE = "header_exchange";public static final String HEADER_QUEUE = "header_queue";@Testpublic void publish()throwsException{//1. 获取连接对象Connection connection = RabbitMQConnectionUtil.getConnection();//2. 构建ChannelChannel channel = connection.createChannel();//3. 构建交换机和队列并基于header的方式绑定channel.exchangeDeclare(HEADER_EXCHANGE, BuiltinExchangeType.HEADERS);channel.queueDeclare(HEADER_QUEUE,true,false,false,null);Map<String,Object> args = new HashMap<>();// 多个header的key-value只要可以匹配上一个就可以// args.put("x-match","any");// 多个header的key-value要求全部匹配上!args.put("x-match","all");args.put("name","jack");args.put("age","23");channel.queueBind(HEADER_QUEUE,HEADER_EXCHANGE,"",args);//4. 发送消息String msg = "header测试消息!";Map<String, Object> headers = new HashMap<>();headers.put("name","jac");headers.put("age","2");AMQP.BasicProperties props = new AMQP.BasicProperties().builder().headers(headers).build();channel.basicPublish(HEADER_EXCHANGE,"",props,msg.getBytes());System.out.println("发送消息成功,header = " + headers);}}