We use Vagrant/Chef as part of our production and development workflows at Conspire. Anyone can trivially spin up a complete dev environment—DB, frontend and backend—on their local machine or log in to our staging machine (an EC2 instance) and interact with our production cluster using Vagrant.
Plus, deploying to production is just a matter of a quick "git pull && ./build.sh && vagrant provision" and checking that everything came back up correctly.
Setting all this up took a long time but it was worth it in the end. At the moment we're just using Chef Solo because we only have four machines to deal with. I imagine we'll move to Chef Server at some point in the future.
I'm unsure how exactly chef does its thing, but I found that using puppet apply (similar to Chef Solo), was far superior to using puppet server (which we use for production).
Maintaining usable certificates, sharing them with other devs, synchronising puppet plugins etc was just far too much work, and very time consuming. I also found it a lot nicer to have the explicit puppet configuration right next to the code that runs on it.
I don't know if these kind of setbacks will be similar with Chef Server or not.
Maybe spend some time looking into puppet then? Although if you've already got everything setup there's probably no need to change. That said, I should probably look into chef a little more than I have.
I've tried looking into salt a couple of times, but I find the documentation really lacking. The information is too spread out. I just took another look at the docs and it appears to do just about everything puppet does, but is much easier to configure (puppet-db/mcollective salt-alternatives are default).
Is there a canonical tutorial that's worth reading?
When I looked at Chef, I decided that it was too much overhead, Chef Solo and its brethren didn't really fix the issue. I found Sprinkle to do what I needed, although the documentation can be confusing.
Plus, deploying to production is just a matter of a quick "git pull && ./build.sh && vagrant provision" and checking that everything came back up correctly.
Setting all this up took a long time but it was worth it in the end. At the moment we're just using Chef Solo because we only have four machines to deal with. I imagine we'll move to Chef Server at some point in the future.