RabbitMQ 消息顺序性保证

方式一：Consumer设置exclusive

注意条件

• 作用于basic.consume
• 不支持quorum queue
  
  当同时有A、B两个消费者调用basic.consume方法消费，并将exclusive设置为true时，第二个消费者会抛出异常：

com.rabbitmq.client.AlreadyClosedException: channel is already closed due to channel error; protocol method: #method<channel.close>(reply-code=403, reply-text=ACCESS_REFUSED - queue 'test' in vhost '/' in exclusive use, class-id=60, method-id=20)at com.rabbitmq.client.impl.AMQChannel.ensureIsOpen(AMQChannel.java:190)at com.rabbitmq.client.impl.AMQChannel.rpc(AMQChannel.java:223)at com.rabbitmq.client.impl.ChannelN.basicConsume(ChannelN.java:981)at com.dms.rabbitmq.TopicSender.lambda$main$2(TopicSender.java:63)at java.base/java.lang.Thread.run(Thread.java:840)

Spring AMQP 如何通过exclusive实现顺序消费：

核心逻辑：

while (!DirectMessageListenerContainer.this.started && isRunning()) {this.cancellationLock.reset();try {for (String queue : queueNames) {consumeFromQueue(queue);}}catch (AmqpConnectException | AmqpIOException e) {long nextBackOff = backOffExecution.nextBackOff();if (nextBackOff < 0 || e.getCause() instanceof AmqpApplicationContextClosedException) {DirectMessageListenerContainer.this.aborted = true;shutdown();this.logger.error("Failed to start container - fatal error or backOffs exhausted",e);this.taskScheduler.schedule(this::stop, Instant.now());break;}this.logger.error("Error creating consumer; retrying in " + nextBackOff, e);doShutdown();try {Thread.sleep(nextBackOff); // NOSONAR}catch (InterruptedException e1) {Thread.currentThread().interrupt();}continue; // initialization failed; try again having rested for backOff-interval}DirectMessageListenerContainer.this.started = true;DirectMessageListenerContainer.this.startedLatch.countDown();
}

1. 抛出异常后，会重试
2. 重试间隔、次数受recoveryInterval（默认无限）、recoveryBackOff控制

方式二：single active consumer

原理：

代码示例：

Channel ch = ...;
Map<String, Object> arguments = newHashMap<String, Object>();
arguments.put("x-single-active-consumer", true);
ch.queueDeclare("my-queue", false, false, false, arguments);

参考资料：https://www.rabbitmq.com/blog/2022/07/05/rabbitmq-3-11-feature-preview-single-active-consumer-for-streams