Structured Ansible Playbooks

Sasha Reid (sasha at- hackl.es) @microwavenby

Just use the PDF of slides: https://bit.do/seagl_ansible https://d.pr/f/Nguxgt

sample flat playbook

create dir

- name:
  file:

clone git repo

- name:
  git:

virtualenv

- name:
  become: no
  pip:
    requirements:

install supervisord

- name:
  apt:
    name: supervisor
    state: latest
- name:
  template:
    src:
    dest:

start app

• it works
• a second one look svery similar and duplicate....

roles & variables

• ansible prioritization of variables is confusing
• roles allow to segment into the required sub-folders
• folders auto-detect main.yml into the folders
• don't duplicate all of the playbook

defaults

what is the difference between variables & defaults? defaults get overridden first, bottom of precedence

basically:

1. group_vars
2. host_vars
3. host facts
4. play vars
5. role vars
6. block vars
7. task vars
8. cli vars

• setting defaults sets documentation of sane

make a role

• make important values into vars

• "role inheritence", make much more generic

• then a "sub-role" that just consumes the generic one!

  • don't set defaults in the sub-role, they smash the meta-role ones

• roles in /meta/ get executed first before /tasks/, and in order

• pass full paths to your meta-roles - they don't get operated in the sub-role context

• only need include_role: tasks for much more control, running multiple times