I'm writing a playbook to setup development environment and build pipeline for a project. I would like to keep playbook decoupled from the actual infrastructure as much as possible, and just rely on groups, so I could setup the pipeline inside a local VM's as well as on the bare hardware with the same command, just using different group limiter.
Now the problem is, with local Vagrant boxes I need to add user vagrant to docker group, where as with bare metal servers some arbitrary user. What is the best practice to handle these kind of variations in the playbook, keeping it as abstract as possible? Should I use behavioral parameters, host variables, group variables, conditionals or facts gathering for this kind of stuff?
Is it even possible or reasonable to keep the playbook fully abstract, without any hardcoded values and adaptability to any environment?
To answer your specific question:
Use a user variable that Vagrantfile overwrites.
For example, it will be ubuntu
if in aws or vagrant
if you are in vagrant.
A more theoretical answer:
i had the same issue with vagrant builds and i decided to keep an ansible playbook somewhere so that all vagrant builds pick it up and execute it to setup the server.
For example, the playbook looks like this:
- hosts: all
tasks:
- name: remove apparmor
apt: name=apparmor purge=true state=absent
and then each of my vagrant repositories wget it and run it with
box.vm.provision :ansible do |ansible|
ansible.playbook = 'vagrant_extra.yml'
ansible.sudo = true
ansible.limit = 'all'
end
The best solution i think would be to use packer to provision your own vagrant image that has all of those things pre-backed in there so that
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With