一、基本概念
1、死信定义
指的是,从队列当中取出来的消息,到达消费方后,因为某些原因导致消息并没有被正常消费掉,这些没有被后续处理的消息就是“死信”,而保存死信的队列,就是死信队列。
2、死信出现的场景举例
为了保证订单业务的消息数据不丢失,需要使用死信队列机制,在消息消费发生异常的时候,将消息给投入到死信队列当中;
用户在商城下单成功并点进去准备支付,超过指定时间未支付时,消息自动失效成为死信(消息超时情况);
3、死信的来源
消息TTL过期;
队列已经到达最大长度,数据无法再添加到MQ;
消息被拒绝了;
4、死信架构图
分析:
对于消息生产者而言,只需要关注将消息发送给交换机即可;
而对于普通消费者C1而言,需要关注普通交换机、普通队列、死信交换机的相关信息,要做两次绑定操作(普通交换机和普通队列绑定,普通队列和死信队列绑定),难点在于---普通队列怎么与死信交换机进行绑定?
而消费者C2也是一个普通消费者,专注于死信队列当中消息的处理,需要关注死信交换机、死信队列的信息,在死信交换机和死信队列绑定之后,从队列当中拿到死信进行处理;
二、代码部分
(0)提前准备工具类
封装获取MQ的connection方法,以及释放资源的方法
public class AMQPUtils {
//用于获取客户端和MQ绑定的连接对象
public static Connection getConnection() throws Exception{
ConnectionFactory factory = new ConnectionFactory();
factory.setVirtualHost("/test");//用于隔离资源的虚拟主机
factory.setHost("MQServer的ip地址");
factory.setPort(5672);
factory.setUsername("zhangsan");
factory.setPassword("1234");
Connection connection = factory.newConnection();
return connection;
}
//释放资源
public static void close(Channel channel,Connection connection) throws Exception{
channel.close();
connection.close();
}
}
(一)模拟消息超时情况
1、消息发布者
public class Provider {
public static void main(String[] args) throws Exception {
Connection connection = AMQPUtils.getConnection();
Channel channel = connection.createChannel();
//声明一个普通交换机
String normalExchange = "normal_Exchange";
channel.exchangeDeclare(normalExchange,"direct");
String key = "zhangsan";
//设置消息的ttl时间为5s
AMQP.BasicProperties properties = new AMQP.BasicProperties().builder().expiration("5000").build();
//发布消息-发布多条消息验证
for (int i = 1; i <= 10 ; i++) {
channel.basicPublish(normalExchange,key,properties,("message---->"+i).getBytes());
}
//释放资源
AMQPUtils.close(channel,connection);
}
}
2、消费者C1
代码:
public class Consumer1 {
public static void main(String[] args) throws Exception{
Connection connection = AMQPUtils.getConnection();
Channel channel = connection.createChannel();
//提前准备一些名字
String normalExchange = "normal_exchange";
String deadExchange = "dead_exchange";
String normalQueue = "normal_queue";
String deadQueue = "dead_queue";
String normal_key = "zhangsan";
String dead_key = "lisi";
//声明普通交换机
channel.exchangeDeclare(normalExchange,"direct");
//声明死信交换机
channel.exchangeDeclare(deadExchange,"direct");
//设置普通队列当中需要携带的其他信息(死信交换机、死信队列、路由key)
Map<String, Object> params = new HashMap<>();
params.put("x-dead-letter-exchange",deadExchange);
params.put("x-dead-letter-routing-key",dead_key);
//声明普通队列
channel.queueDeclare(normalQueue,false,false,false,params);
//声明死信队列
channel.queueDeclare(deadQueue,false,false,false,null);
//binding
//将普通交换机和普通队列绑定
channel.queueBind(normalQueue,normalExchange,normal_key);
//将将死信交换机和死信队列绑定
channel.queueBind(deadQueue,deadExchange,dead_key);
//消费消息
channel.basicConsume(normalQueue,true,new DefaultConsumer(channel){
@Override
public void handleDelivery(String consumerTag, Envelope envelope, AMQP.BasicProperties properties, byte[] body) throws IOException {
System.out.println("C1消费的消息是---->"+new String(body));
}
});
}
}
注意:开启C1接受普通交换机的消息之后,关闭C1,让普通队列当中的消息超过超时时间,成为死信,后被死信交换机路由进入dead_queue当中,下图所示的就是消息超时之后进入死信队列:
点击dead_queue可以查看具体的死信来源、交换机、路由key等信息;
此时:注意此时死信消息是保存在MQ当中的;
3、消费者c2
消费者C2要去消费死信队列当中的消息:
public class Consumer2 {
public static void main(String[] args)throws Exception {
Connection connection = AMQPUtils.getConnection();
Channel channel = connection.createChannel();
//准备一些名字
String deadExchange = "dead_exchange";
String deadQueue = "dead_queue";
String key = "lisi";
//声明死信交换机
channel.exchangeDeclare(deadExchange,"direct");
//声明死信队列
channel.queueDeclare(deadQueue,false,false,false,null);
//交换机和队列绑定
channel.queueBind(deadQueue,deadExchange,key);
//消费死信消息
channel.basicConsume(deadQueue,true,new DefaultConsumer(channel){
@Override
public void handleDelivery(String consumerTag, Envelope envelope, AMQP.BasicProperties properties, byte[] body) throws IOException {
System.out.println("C2消费了死信----->"+new String(body));
}
});
}
}
控制台结果:
此时存储在死信队列当中的消息已经被C2消费了!
(二)模拟队列达到最大长度
请提前在MQ的控制台上,将情况1当中设置的队列给删除;
1、消息发布者
public class Provider {
public static void main(String[] args) throws Exception {
Connection connection = AMQPUtils.getConnection();
Channel channel = connection.createChannel();
//声明一个普通交换机
String normalExchange = "normal_exchange";
channel.exchangeDeclare(normalExchange,"direct");
String key = "zhangsan";
//发布消息-发布多条消息验证
for (int i = 1; i <= 10 ; i++) {
channel.basicPublish(normalExchange,key,null,("message---->"+i).getBytes());
}
//释放资源
AMQPUtils.close(channel,connection);
}
}
2、消费者C1
public class Consumer1 {
public static void main(String[] args) throws Exception{
Connection connection = AMQPUtils.getConnection();
Channel channel = connection.createChannel();
//提前准备一些名字
String normalExchange = "normal_exchange";
String deadExchange = "dead_exchange";
String normalQueue = "normal_queue";
String deadQueue = "dead_queue";
String normal_key = "zhangsan";
String dead_key = "lisi";
//声明普通交换机
channel.exchangeDeclare(normalExchange,"direct");
//声明死信交换机
channel.exchangeDeclare(deadExchange,"direct");
//设置普通队列当中需要携带的其他信息(死信交换机、死信队列、路由key)
Map<String, Object> params = new HashMap<>();
params.put("x-dead-letter-exchange",deadExchange);
params.put("x-dead-letter-routing-key",dead_key);
//设置普通队列的最大长度
params.put("x-max-length",6);
//声明普通队列
channel.queueDeclare(normalQueue,false,false,false,params);
//声明死信队列
channel.queueDeclare(deadQueue,false,false,false,null);
//binding
//将普通交换机和普通队列绑定
channel.queueBind(normalQueue,normalExchange,normal_key);
//将将死信交换机和死信队列绑定
channel.queueBind(deadQueue,deadExchange,dead_key);
//消费消息
channel.basicConsume(normalQueue,true,new DefaultConsumer(channel){
@Override
public void handleDelivery(String consumerTag, Envelope envelope, AMQP.BasicProperties properties, byte[] body) throws IOException {
System.out.println("C1消费的消息是---->"+new String(body));
}
});
}
}
开启C1,然后关闭,再启动消息生产者,结果:
3、消费者C2
public class Consumer2 {
public static void main(String[] args)throws Exception {
Connection connection = AMQPUtils.getConnection();
Channel channel = connection.createChannel();
//准备一些名字
String deadExchange = "dead_exchange";
String deadQueue = "dead_queue";
String key = "lisi";
//声明死信交换机
channel.exchangeDeclare(deadExchange,"direct");
//声明死信队列
channel.queueDeclare(deadQueue,false,false,false,null);
//交换机和队列绑定
channel.queueBind(deadQueue,deadExchange,key);
//消费死信消息
channel.basicConsume(deadQueue,true,new DefaultConsumer(channel){
@Override
public void handleDelivery(String consumerTag, Envelope envelope, AMQP.BasicProperties properties, byte[] body) throws IOException {
System.out.println("C2消费了死信----->"+new String(body));
}
});
}
}
结果:
死信队列当中的消息已经被消费了;
(三)模拟消息被拒绝
1、消息发布者
public class Provider {
public static void main(String[] args) throws Exception {
Connection connection = AMQPUtils.getConnection();
Channel channel = connection.createChannel();
//声明一个普通交换机
String normalExchange = "normal_exchange";
channel.exchangeDeclare(normalExchange,"direct");
String key = "zhangsan";
//发布消息-发布多条消息验证
for (int i = 1; i <= 10 ; i++) {
channel.basicPublish(normalExchange,key,null,("message---->"+i).getBytes());
}
//释放资源
AMQPUtils.close(channel,connection);
}
}
2、消费者C1
public class Consumer1 {
public static void main(String[] args) throws Exception{
Connection connection = AMQPUtils.getConnection();
final Channel channel = connection.createChannel();
//提前准备一些名字
String normalExchange = "normal_exchange";
String deadExchange = "dead_exchange";
String normalQueue = "normal_queue";
String deadQueue = "dead_queue";
String normal_key = "zhangsan";
String dead_key = "lisi";
//声明普通交换机
channel.exchangeDeclare(normalExchange,"direct");
//声明死信交换机
channel.exchangeDeclare(deadExchange,"direct");
//设置普通队列当中需要携带的其他信息(死信交换机、死信队列、路由key)
Map<String, Object> params = new HashMap<>();
params.put("x-dead-letter-exchange",deadExchange);
params.put("x-dead-letter-routing-key",dead_key);
//声明普通队列
channel.queueDeclare(normalQueue,false,false,false,params);
//声明死信队列
channel.queueDeclare(deadQueue,false,false,false,null);
//binding
//将普通交换机和普通队列绑定
channel.queueBind(normalQueue,normalExchange,normal_key);
//将将死信交换机和死信队列绑定
channel.queueBind(deadQueue,deadExchange,dead_key);
//消费消息--注意关闭自动确认
channel.basicConsume(normalQueue,false,new DefaultConsumer(channel){
@Override
public void handleDelivery(String consumerTag, Envelope envelope, AMQP.BasicProperties properties, byte[] body) throws IOException {
/*System.out.println("C1消费的消息是---->"+new String(body));*/
//模拟消息被拒绝--把所有消息都拒绝
channel.basicReject(envelope.getDeliveryTag(),false);
}
});
}
}
注意:先启动C1,然后关闭,启动消息发布者,结果如下:
这10条消息目前都保存在MQ当中,然后再启动C1,把消息全部拒绝掉,让它们成为死信:
点击dead_queue,去查看死信队列当中的一些信息:
3、消费者C2
public class Consumer2 {
public static void main(String[] args)throws Exception {
Connection connection = AMQPUtils.getConnection();
Channel channel = connection.createChannel();
//准备一些名字
String deadExchange = "dead_exchange";
String deadQueue = "dead_queue";
String key = "lisi";
//声明死信交换机
channel.exchangeDeclare(deadExchange,"direct");
//声明死信队列
channel.queueDeclare(deadQueue,false,false,false,null);
//交换机和队列绑定
channel.queueBind(deadQueue,deadExchange,key);
//消费死信消息
channel.basicConsume(deadQueue,true,new DefaultConsumer(channel){
@Override
public void handleDelivery(String consumerTag, Envelope envelope, AMQP.BasicProperties properties, byte[] body) throws IOException {
System.out.println("C2消费了死信----->"+new String(body));
}
});
}
}
启动消费者C2,消费死信队列当中的消息!
三、小结
死信队列的出现,是为了保存因为特殊原因无法被消费的消息,避免消息直接失效!这些消息通过rabbitMQ的死信队列机制,可以保存在MQ服务的死信队列当中,等待被其他的消费者进行处理!
需要注意的是:
只有针对消息的设置会放在消息发布方进行,队列等操作,因为发布方无法自己决定消息被路由到哪个队列,只能决定把消息交给哪个交换机,以及给定路由规则;
对于消息消费方而言,需要确定交换机、消息队列,已经完成 交换机和队列的绑定操作,所以针对于队列的设置都是放在消费方完成的!
版权归原作者 山野万万里! 所有, 如有侵权,请联系我们删除。