Single RabbitMQ queue and multiple routing key

Question

We've got an application that will be using RabbitMQ. The design is to use a single exchange single queue with multiple routing keys for multiple teams and they will communicate through this single queue. I'm developing a java Application to just listen to that queue using a routingKey assigned to my team.

    @RabbitListener(bindings = @QueueBinding(
            value = @Queue(value = "queue", durable = "true"),
            exchange = @Exchange(value = "exchange", autoDelete = "false", type = "topic"),
            key = "abc_rk"))
    public void consumeMessagesFromRabbitMQ(Request request) throws InterruptedException {
        System.out.println("Start:Request from RabbitMQ: " + request);
        Thread.sleep(10000L);
        System.out.println("End:Request from RabbitMQ: " + request);
    }

Let's say the queue has 3 routingKey messages and out of them my application just want to listen to abc_rk. But when I run this code, it's not filtering out the other messages but instead irrespective of what I've set in "key = ?" it pulls all the messages from the queue. Note that I can't change the design and use the separate queue for each routingKey.

score 1 · Accepted Answer · answered Jul 13 '22 at 12:56

RabbitMQ doesn't work that way (it has no concept of a message selector unlike JMS).

In fact, consumers know nothing about routing keys, only producers; the only reason you see it on @RabbitListener is to aid configuration.

To do what you want, you need to bind 3 different queues to the exchange with the respective routing keys.

Note that I can't change the design and use the separate queue for each routingKey.

You could add a MessagePostProcessor to the container (afterReceivePostProcessors) to discard the unwanted messages by returning null. That is the only mechanism the framework provides for filtering messages.

/**
 * Set {@link MessagePostProcessor}s that will be applied after message reception, before
 * invoking the {@link MessageListener}. Often used to decompress data.  Processors are invoked in order,
 * depending on {@code PriorityOrder}, {@code Order} and finally unordered.
 * @param afterReceivePostProcessors the post processor.
 * @since 1.4.2
 * @see #addAfterReceivePostProcessors(MessagePostProcessor...)
 */
public void setAfterReceivePostProcessors(MessagePostProcessor... afterReceivePostProcessors) {

But the best solution is 3 queues.

Single RabbitMQ queue and multiple routing key

1 Answers1