Spring Cloud Stream RabbitMQ leave message on dead letter queue

Question

Im using Spring cloud streams to consume message from rabbitmq. Im trying to get a failed message to stay on a dead letter queue after a number of retries. I have programmatically done this before using amqp but it seems to be a bit more difficult to get right with spring cloud streams.

@StreamListener(target = Sink.INPUT)
public void messageListener(final String in, @Header(name = "x-death", required = false) Map<?, ?> death) {

  if (!validString(in)) {
    // We don’t need this message anymore not even on the dlq
    throw new ImmediateAcknowledgeAmqpException(“String not good”);
  }

  // If message has been retried more than 3 time we want to put message on dlq
  if (death != null && death.get("count").equals(3L)) {
    // I know this is incorrect as it will ack the message, but at this point I need the message to be left on the dlq
    throw new ImmediateAcknowledgeAmqpException("Failed after 4 attempts");
  }

  try {
    // this trows an exception
    processService.process(in);
  }
  catch (Exception ex) {
    // here we retry the message
    throw new AmqpRejectAndDontRequeueException("retry message");
  }
}

My Configuration

spring.cloud.stream.bindings.input.destination=adestination
spring.cloud.stream.bindings.input.group=aqueue
spring.cloud.stream.rabbit.bindings.input.consumer.bindingRoutingKey=akey
#dlx/dlq setup
spring.cloud.stream.rabbit.bindings.input.consumer.deadLetterQueueName=adeadletterqueue
spring.cloud.stream.rabbit.bindings.input.consumer.dlqDeadLetterExchange=
spring.cloud.stream.rabbit.bindings.input.consumer.autoBindDlq=true
spring.cloud.stream.rabbit.bindings.input.consumer.requeueRejected=true
spring.cloud.stream.rabbit.bindings.input.consumer.dlqTtl=5000
# disable binder retries
spring.cloud.stream.bindings.input.consumer.max-attempts=1

Any help greatly appreciated

Thanks

Answer 1

You can't modify a rejected message (e.g. adding a TTL header).

I believe you will have to manually publish to a parking-lot queue when the retries are exhausted.