[rabbitmq-discuss] Clustering of Rabbit MQ on thwo physical boxes.

Simon MacMullen simon at rabbitmq.com
Wed Nov 16 13:33:50 GMT 2011

On 16/11/11 12:00, GS wrote:
> I had a question about clustering rabbit on two separate physical
> machine. The OS is unix. I was going through some blogs like
> http://www.godlikemouse.com/2010/12/14/how-to-cluster-rabbit-mq/ and
> it says that you have to change the /etc/hosts file. My organization
> does not permits that change. IS it still possible to have a cluster
> setup on two separate physical machines, without changing the /etc/
> hosts file?

Yes. That blog seems slightly confused. The point is that RabbitMQ will 
identify each cluster node based on its short (non-fully-qualified) 
hostname, and thus these hostnames are expected to resolve to the 
correct IP addresses. (I would argue that if they don't then your 
network is broken.) The blog author was presumably on a network where 
this was not the case, and used /etc/hosts hacking to fix it.

> Are the steps mentioned at http://www.rabbitmq.com/clustering.html
> sufficient?

They certainly should be.

> I am not doubting the content, but my experience of
> http://www.rabbitmq.com/clustering.html#single-machine was not good.

Hmm. Clustering on a single machine is more painful: you have to make 
sure that the multiple copies of RabbitMQ that you're running don't 
clash on any ports, store data in different locations, etc. That 
requires a bit of fiddling. It's still possible but it's easier to get 
something wrong and thus not even be able to start all your nodes.

Cheers, Simon

Simon MacMullen
RabbitMQ, VMware

More information about the rabbitmq-discuss mailing list