We're working on solving these sorts of devops problems in Ubuntu with Juju: https://juju.ubuntu.com/
As an alternative approach would be that a Juju charm (script) would handle the initial deployment of a stock Ubuntu AMI and the customization in one step (or with puppet/chef) and then allow you to add new instances based on scale (though currently not automatic). When you have changes to your service you update the charm and just `juju upgrade-charm`.
This would consolidate step 1, 5, and the "ongoing" into one tool and you'd get a cloud-agnostic deployment (openstack cloud or bare-metal).
I would really love to recommend Juju. I used it for a good couple months attempting to get even just a working Openstack deployment. The trouble is that for the bare-metal functionality you guys mean it to be paired with MaaS. Unfortunately the community I found to exist around this combination was basically zero.
Even a simple question went unanswered the couple times I posted it. Eventually I gave up and just went back to Puppet+Openstack modules for the configuration I was working on at the time.
This was exactly my experience as well. Then I was introduced to Mirantis Fuel which uses Cobbler/Puppet to do bare metal provisioning and remote configuration.
Very cool. I looked at the Fuel stuff as well, but at the time it wasn't quite able to do everything I needed so I ended up rolling my own. The 2.2 update they have coming up soon sounds like it fixes all of the things that were 'wrong' for me though!
FWIW, I tested juju a few months ago and found it to be buggy and unreliable. Sometimes the instances would connect together correctly, and sometimes they would fail inexplicably. Didn't seem ready for any kind of production use to replace config mgmt tools.
I have tried ansible and love it. I invested the time (a few hours) to get a script working for my stack, and now I have a 70ish line script that will provision a server or VM (with a shared code directory in the VM), from clean installation to the codebase up and running in its production state, in one command.
New project? I just copy the script, change a few variables (names and packages it needs), and I get deployment for free.
Sounds interesting. Any possibility that you could anonymize the script and paste a link to a gist of it? would be interesting to see some real-world Ansible examples.
I never thought of Ansible for orchestration particularly, but that sounds interesting! The current orchestration support in saltstack is somewhat limited, so it would be neat to check out.
Are people using Ansible to replace or manage fabric scripts? I'd like to figure out some way to limit the spread of one-off fabric scripts (reminds me too much of bash script proliferation) and it'd be great to get everything under one roof.
As an alternative approach would be that a Juju charm (script) would handle the initial deployment of a stock Ubuntu AMI and the customization in one step (or with puppet/chef) and then allow you to add new instances based on scale (though currently not automatic). When you have changes to your service you update the charm and just `juju upgrade-charm`.
This would consolidate step 1, 5, and the "ongoing" into one tool and you'd get a cloud-agnostic deployment (openstack cloud or bare-metal).