in network, python, rabbitmq

Default behaviour in implementation of STOMP protocol in RabbitMQ with python


Why STOMP, and not directly AMQP, as I’m using RabbitMQ. No real reason, but the fact that there are less dependencies on a STOMP client, as it’s just a socket with text sent.


There are several implementations of the STOMP protocol for Python. The module I chose is python-stomp (version 0.2.9), from Benjamin W. Smith. It’s simple and easy to understand.

Simple Code Examples

Everything is working fine, when launching, I receive the message. But when I launched several receivers, I noticed, that ONLY ONE programs received the message! After some research, I found the answer: As documented in the RabbitMQ wiki, the default exchange is ‘direct’:

[…]when messages leave a queue for a consumer, they are not duplicated. One message, sitting on a queue, is delivered to only one of the available consumers. […] If there are multiple clients, all SUBSCRIBEing to the same queue, then there will be multiple consumers all on the same queue, leading to round-robin delivery to those clients.

There is an explanation on how to change the behaviour, by changing the exchange type, and some of particular bits (like the id). I even found an example of modification for use in the equivalent STOMP Ruby module.

Here are the modifications. The good news is that there is no need to patch the stompy module, as the author provided the possibility to pass arbitrary parameters to the headers by the use of the ‘conf’ variable.

The important points are:

  • You need to define an exchange of type amq.topic
  • You need to set an id, which is different for each client
  • As you’re using topics, you’ll have to specify a routing_key

  1. @lstep @monadic Wow, that was quick! He just emailed me this morning asking about that! #STOMP

Comments are closed.


  • Anonymous August 27, 2010

    Twitter Trackbacks…