Skip to content


Subversion checkout URL

You can clone with
Download ZIP
Generate CloudInit scripts for multiple server roles based on a collection of files.
Fetching latest commit...
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.

EC2 Rolegen

How to use it

On EC2 images that use cloud-init, this can generate "user data scripts" for bootstrapping systems to use different roles. This allows users to configure the systems without needing to put themselves through the tedious process of creating custom AMIs (especially in consideration of micro instances) and allows you to keep your user data scripts in a versioned repository - if desired - with a simple tool that automates the process of generating the scripts.

This script will leverage CloudInit's ability to use multi-part MIME files in order to allow any number of scripts to coexist, and gzips the user data scripts.

A simple idea of using this might be that you want a server, but you don't want to make a custom AMI and you don't want to manually configure it. You could create a directory hierarchy next to like this:

  |- example/

Now, you could have the following inside of your file:

echo 'Hello, world.' > /root/hello.txt

Now, execute this command - and you will have a final packge called example.gz in the ./dist directory:


How it works

This is pretty simple. All that this script does is look for a directory called 'roles' in your current working directory. If this directory is not found, then it will be created. It will then search this directory for any files or directories.

If the script finds a directory, then it creates a new role which is given the same name as that directory. If it finds a file, then that file is added to a list of files that should be used for every role. These are global files.

For each directory, the script collects all global files and all files in that roles directory and puts them into a gzipped (compression level 9) mime file in the dist directory. These files can be used as user data for your instance in order to bootstrap it.


Something went wrong with that request. Please try again.