Clustering with standalone docker

Archives AWX Project
1

Hi AWX community,

Did anyone get any success with clustering up some standalone docker instances? I’ve followed the clustering guide from Tower 3.2.1 documentation but I can’t get to a state where the two instances I have are aware of each other at all.
I’ve seen a couple of posts here about dynamic clustering being worked on only for OpenShift (1), instance groups worked on for OpenShift again (2) and the instance groups being “technically supported but hard to configure in AWX” (3).

What I need is a redundant setup of AWX machines that doesn’t multiplicate jobs (preferably with some “jump hosts”, but that’s not required). Do you have any recommendations on that?

(1) https://github.com/ansible/awx/issues/74
(2) https://groups.google.com/d/topic/awx-project/_7PQp8PD96g/discussion
(3) https://groups.google.com/forum/#!searchin/awx-project/clustering|sort:date/awx-project/wI5LCwNBYaA/QGfFyUmlAgAJ

Best regards,
Piotr

2

This is unlikely to be directly supported by us any time in the near future. Unless someone contributes installer roles to support something like docker swarm.

3

Hi Matthew,

Many thanks for coming back to me on that. Given RedHat priority to run on OpenShift, this doesn’t surprise me much. I think we’ll have to run a standby for now and look into the Docker Swarm deployment when we have some cycles.

I understand that the Tower clustering is based mainly on RabbitMQ, but do you have any design docs or articles on how it actually works? I’d need more study material if I was to configure this :slight_smile:

Best regards,
Piotr

4

Hi Piotr - Have you look into using docker swarm yet ?