Introduction

Describe your infrastructure in YAML like this:

---
droplets:
  - name: development.web.1
  - name: development.app.1
  - name: development.db.1
  - name: production.web.1
    region: 2
  - name: production.web.2
  - name: production.app.1
    region: 2
  - name: production.app.2
  - name: production.db.1
    size: 65

Run an ansible playbook to create droplets on Digital Ocean, and then run ansible to execute remote commands across groups of newly created machines.

Purpose

This repository is mainly an exploration of Ansibles provisioning and orchestration capabilities. It's tied to using the Digital Ocean cloud provider just to make testing easier, but in principle could be extended or altered to support other providers.

It allows you to define a set of droplets in a file, then use Ansible to create those instances. Once created Ansible uses the Digital Ocean API to discover the addresses and for you to run commands across portions of your new infrastructure. Because Digital Ocean doesn't yet support any sort of metadata on droplets the name is used to encode information.

Installation

After cloning this repository you need to install Ansible and a couple of dependencies.

pip install -r requirements.txt

Note that this currently uses a pre-release version of Ansible which provides an improved version of the Digital Ocean module.

You also need to set a couple of environment variables for the Digital Ocean API.

export DO_API_KEY=xxxxxxxxxxxxxxx
export DO_CLIENT_ID=xxxxxxxxxxxxxxxxxx

Usage

The repository contains example files for you to try out but the format is quiet simple. First up lets use a simple example that creates a single droplet (remember this requires you to have a Digital Ocean account and will cost you money):

cp vars/small.yml.example vars/droplets.yml
cp vars/custom.yml.example vars/custom.yml
cp host_vars/localhost.example host_vars/localhost

You'll now need to edit vars/custom.yml and enter the numberic id of your ssh key. This is needlessly fiddly, I would recommend first installing the Tugboat client and then running:

tugboat keys

With that done we can idempotently create the machines specified in the above file simply by running the playbook. Note that the first run will take a few seconds while it hits the Digital Ocean API but these results are cached.

ansible-playbook -i hosts provision_digital_ocean.yml

Now you have some machines up and running you can use the ansible tool to list the hosts, or to run arbitrary commands on them like so.

ansible -i hosts all --list-hosts
ansible -i hosts all -a 'uptime'

Groups

More useful is if you adopt a strong naming convention for your droplets. Note the following template:

<environment>.<type>.<id>

Using this format will generate Ansible groups for all environments used and for all combinations of environment/type. For instance the above example configuration will create groups for:

• production
• development
• development_web
• development_app
• development_db
• production_web
• production_app
• production_db

This means you can direct commands easily to the relevant droplets like so:

ansible -i hosts production_web -a 'uptime'

Advanced

Default values for the region, size and base image are provided in the host_vars/localhost.example file. This should be copied to host_vars/localhost from where it can be edited. You can override the provided values with your own values as well as specify these on a per droplet basis if required.

The following places two droplets in Amsterdam and increases the size of the production.db.1 droplet to 8GB.

---
droplets:
  - name: development.web.1
  - name: development.app.1
  - name: development.db.1
  - name: production.web.1
    region: 2
  - name: production.web.2
  - name: production.app.1
    region: 2
  - name: production.app.2
  - name: production.db.1
    size: 65

To get at the numeric ids you need for this again I'd recommend the Tugboat client mentioned above.