There is quite some documentation available already, but it wouldn't hurt to briefly explain how to use these roles.
- The roles work on as many distribution as possible. The intent is that you can switch the operating system, without changing your playbook or variables. There are exceptions, for example Zabbix is only provided for a few operating systems.
- These roles are as simple as possible. There can be cases where complicated stuff happens, but that's required from time to time.
- These roles are thoroughly tested. On each commit, pull request and release and also on full virtual machines.
Before using any role, they need to be installed. There are a few ways to do that:
- Install them one-by-one using
ansible-galaxy install buluma.bootstrap
for example. - Install them using a
requirements.yml
file (see below) usingansible-galaxy install --role-file requirements.yml
.
The requirements.yml
file is a list of roles, for example:
---
- buluma.bootstrap
These roles are typically installed in ~/.ansible/roles/
. You can configure Ansible to look somewhere else though, for example this ansible.cfg
stores the roles "locally", in the same directory where the ansible.cfg is found:
[default]
roles_path = roles
Once the roles are downloaded, you can integrate them into a playbook, for example:
---
- name: bootstrap all hosts
hosts: all
become: yes
gather_facts: no
roles:
- role: buluma.bootstrap
You can also use [include_role](https://docs.ansible.com/ansible/latest/modules/include_role_module.html)
.
Most roles can be controlled using variables. For example by using the vars:
statement:
---
- name: bootstrap all hosts
hosts: all
become: yes
gather_facts: no
roles:
- role: buluma.bootstrap
bootstrap_user: root
Most roles have quite a few variables and the playbook can get long using the example above. Another way to use variables in in the inventory, specifically in the group variables. For example the file group_vars/all.yml
can contain:
---
bootstrap_user: root
This is a layout of the file and directory structure that I tend to use:
├── ansible.cfg
├── inventory
│ ├── group_vars
│ │ └── all.yml
│ └── hosts
├── playbook.yml
└── roles
├── requirements.yml
└── buluma.bootstrap
Also see best practices.
[defaults]
roles_path=roles
retry_files_enabled=no
inventory=inventory
Also see Ansible configuration.
fedora-s-1vcpu-2gb-fra1-01 ansible_host=167.99.141.134
Also see working with inventories.
---
bootstrap_wait_for_host: yes
Also see group variables
---
- name: bootstrap all hosts
hosts: all
become: yes
gather_facts: no
roles:
- buluma.bootstrap
Also see working with playbooks.
---
- buluma.bootstrap
Also see installing roles from a file.
Although a consumer of the role does not need to look into the role and ansible-galaxy install
will manage the contents of these roles, here is a brief explanation what can be found here:
defaults/main.yml
- Variables that can be overwritten by you, the consumer of the role.tasks/main.yml
- A list of tasks that this role executes.vars/main.yml
- Variables that should not be overwritten by you. For example a mapping from operating system and the packages that should be installed.templates/*
- Files that are generated on the managed node, these typically contain variable that are substituded for a value. The files in here should be mentioned in thetasks/main.yml
.molecule/*
- A few scenarios that the role is tested against. I typically see a scenario as a distribution.files/*
- A set of static files. The files in here should be mentioned in thetasks/main.yml
.handlers/main.yml
- Tasks that are executed after the play when called from a task intasks/main.yml
when the task is changed.meta/main.yml
- Descriptive information about the role.README.md
- Documentation how to use the role.
Also see Ansible roles.