Planning thoughts...

[brianfeister]

@patcon and @jeffreynolte - Patrick has alot of experience with DevOps stuff (something that I'm more new to). Jeffrey is another dev looking for a good LEMP box that accommodates WordPress. @santiagogil is the owner of the repo (who has also offered some tips) that I forked in creating this.

My goal is to have a generic environment for WordPress, but it will be geared toward WP-Multisite which I am using for my project. If someone wants to maintain a WP solo instance, then you'll want to fork this and maintain some alternate options in a branch. Some of the specs for what I'm working toward:

• At the recommendation of @evansolomon on his blog about an nginx stack for WP multisite I'm planning to using Debian Squeeze - hopefully the Linux distro is not an issue and the chef recipes work well regardless of the Linux distro specified in the config.
• I want to incorporate both a Vagrant box and also a production strategy with knife-solo so that a new server can be provisioned with a mirror config of the vagrant box.
• The base repo will include WP-Multisite (and do setup for it)

@patcon told me in a side conversation that he uses rake to glue setup bits together. Curious if this is something that knife-solo aims to do or if I'm overlooking something in jumping to that conclusion. He referenced this rakefile https://github.com/myplanetdigital/jenkins-inception/blob/develop/Rakefile#L18-L48 from his day job and right now I'm trying to figure out if there are pro's / con's or if I'm just missing something obvious in deciding between using a rakefile or using knife-solo for the production strategy. I'm not very fluent in Ruby and have not written a rakefile. I thought that chef accepted JSON (I've got more background in javascript and JSON is comfortable) style parameters for the various server config aspects of the build process... am I wrong there?

[brianfeister]

When I talk about "production strategy" I'm referring to what might be better called "provisioning". Just to clarify in case that wasn't clear.

[evansolomon]

Glad to hear my post was useful. There are some updates I'd like to make to it at some point, but it should be a good starting point.

Good luck with the project.

[brianfeister]

Thanks @evansolomon, your walkthrough was a tremendous help and got me started on this whole thing in the first place. Just wanted to give credit where due, since I'm valuing your opinion highly with regard to server config, etc.

[patcon]

Haven't caught up yet in the above reading, but I would recommend that you consider server setup as a job for Chef (installing and configuring all the packages), and deployment of the wordpress site as a job for a dedicated deployment tool like capistrano. I use rake now, but the plan was always to move toward a real deployment tool, and take the application-specific things out of Chef's hands.

This approach would allow the LEMP project to be used more widely, as it's agnostic about the app (Wordpress, Drupal, ImpressCMS, whatever). Would also make deployment to the remote server and the local vagrant box identical :)

[patcon]

He referenced this rakefile https://github.com/myplanetdigital/jenkins-inception/blob/develop/Rakefile#L18-L48 from his day job and right now I'm trying to figure out if there are pro's / con's or if I'm just missing something obvious in deciding between using a rakefile or using knife-solo for the production strategy

Oh hey, and to clear up the confusion, I use rake to spin up the os image of the server itself (with Rackspace using the fog gem), since knife-solo doesn't actually create a new cloud server for me -- knife-solo just makes it simple to run chef-solo on the fresh remote server after it exists :)