README.md

coreos-bootstrap

In order to effectively run Ansible, the target machine needs to have a Python interpreter. CoreOS machines are minimal and do not ship with any version of Python. To get around this limitation we can install PyPy, a lightweight Python interpreter. The coreos-bootstrap role will install PyPy for us and we will update our inventory file to use the installed python interpreter.

Install

Add to your requirements.yml:

- name: instrumentisto.coreos-bootstrap
  src: git+https://github.com/instrumentisto/ansible-coreos-bootstrap
  version: master

And resolve your dependencies:

ansible-galaxy install -r requirements.yml

Or add the role directly:

ansible-galaxy install git+https://github.com/instrumentisto/ansible-coreos-bootstrap

Configure your project

Unlike a typical role, you need to configure Ansible to use an alternative Python interpreter for CoreOS hosts. This can be done by adding a coreos group to your inventory file and setting the group's vars to use the new Python interpreter. This way, you can use Ansible to manage CoreOS and non-CoreOS hosts. Simply put every host that has CoreOS into the coreos inventory group and it will automatically use the specified Python interpreter.

[coreos]
host-01
host-02

The role sets defaults for these, but you can set them yourself just in case:

[coreos:vars]
ansible_ssh_user=core
ansible_python_interpreter=/opt/python/bin/python

This will configure Ansible to use the Python interpreter at /opt/python/bin/python which will be created by the coreos-bootstrap role.

Bootstrap Playbook

Now you can simply add the following to your playbook file and include it in your site.yml so that it runs on all hosts in the coreos group.

- hosts: coreos
  gather_facts: False
  become: yes
  roles:
    - instrumentisto.coreos-bootstrap

Make sure that gather_facts is set to False, otherwise Ansible will try to first gather system facts using Python which is not yet installed!

Example Playbook

After bootstrap, you can use Ansible as usual to manage system services, install Python modules (via pip), and run containers. Below is a basic example that starts the etcd service, installs the docker-py module and then uses the Ansible docker module to pull and start a basic Nginx container.

- name: Nginx Example
  hosts: web
  become_method: sudo
  become: yes
  tasks:
    - name: Start etcd
      systemd:
        name: etcd.service
        state: started

    - name: Install docker-py
      pip:
        name: docker-py
        state: present
        executable: /opt/python/bin/pip

    - name: Pull container
      raw: docker pull nginx:1.7.1

    - name: Launch Nginx container
      docker:
        name: example-nginx
        image: nginx:1.7.1
        ports: "8080:80"
        state: running

License

MIT