After reading the docs it seems like the supported way for doing clustering is to just add your hosts in the inventory file and rerun the install playbook which then install AWX on all the included hosts. The confusion I have is that a single AWX instance includes nginx and rabbitmq so if I have a cluster, do I then have nginx and rabbitmq running on each instance? I’m used to the architecture of having a single load balancer and a single instance of rabbitmq with multiple instances of the web service behind it.
Currently clustering only supports openshift and is not yet available in the devel branch. You can find the work for it in the scalable_clusters branch.
There are also several posts in this mailing list where I go over the architecture and reasoning behind the cluster implementation.