wp:Ansible_(software) is an open-source software platform for configuring and managing computers. It combines multi-node software deployment, ad hoc task execution, and configuration management. Written in Python, it is packaged by RedHat. As of July 2016, we're using Ansible 2.2.0
A lot of this page is outdated.
Also, RedHat seems to have purposely made things very convoluted in terms of versioning, release cycles and product naming. So check https://docs.ansible.com/ansible/latest/reference_appendices/release_and_maintenance.html and see if you can figure it out.
Contents
- 1 Installation
- 2 Getting Started
- 3 Modules
- 4 Example Commands
- 5 Variables
- 6 Playbooks
- 7 Targets
- 8 Testing
- 9 Best Practices
- 10 Scope
- 11 Ansible with Vagrant
- 12 Ansible with MediaWiki
- 13 Ansible with Drupal
- 14 Ansible in the cloud
- 15 Ansible on Fedora
- 16 Ansible Docs
- 17 Ansible References
- 18 References
Installation
The preferred way to install is to just git clone
the source. Having the source makes it easy to upgrade, and it's self-contained, plus best of all you get all the examples and contribs.
cd
mkdir ~/bin
cd bin
git clone git://github.com/ansible/ansible.git --recursive
cd ./ansible
source ./hacking/env-setup
If you see this error message:
Traceback (most recent call last): File "/usr/local/bin/ansible-playbook", line 44, in <module> import ansible.constants as C ImportError: No module named ansible.constants
Be sure to source the env-setup script
Getting Started
You must source the environment setup script to begin using Ansible (assuming you are running from a git checkout) source ~/bin/ansible/hacking/env-setup
Ansible provides three main commands:
ansible-playbook
- to execute an Ansible playbook on the specified systemsansible
- to execute an individual shell command or Ansible module on the specified systemsansible-vault
- (optional) to encrypt or decrypt YAML files that Ansible uses.
Global Config
export ANSIBLE_HOST_KEY_CHECKING=False
or set it in your ~/.ansible.cfg so that as you add new hosts it won't prompt you.
Also use 'ssh' instead of paramiko when doing this.
Initialize a Project
Use Ansible Galaxy If you want to do a new project, you can use the ansible-galaxy foo init
command which will create the directory and file structure for 'foo' in the current working directory.
Also, if you want to install other Ansible Galaxy projects, you can either do it "manually" ansible-galaxy install -r
Or, setup a 'requirements.yml' file in your playbook that then gets run by your stack. [1][2]
ansible-galaxy [delete|import|info|init|install|list|login|remove|search|setup] [--help] [options]
Modules
Ansible comes with over 200 modules that you should get familiar with in order to use the system effectively.
There are a bunch of modules in Ansible, like the MySQL module, the Monit module, or the File module and other interesting modules like jabber, mail, sendgrid, dpkg_selections, composer, yum, redhat_subscription, digital ocean, the authorized_key module for working with SSH keys, and a whole section of system modules.
You can use the command module (secure but simple) or the shell module. The latter may be useful if you need to run bash explicitly (defaults to /bin/sh); or anytime you need $HOME and redirection.
To sanitize any variables passed to the shell module, you should use "{{ var | quote }}" instead of just "{{ var }}" to make sure they don't include evil things like semicolons.
Example Commands
Note: control verbosity with -vvvv
ansible --help
display helpansible --version
show version infoansible -c local -i ~/ansible_hosts -m ping all
ping all the hosts in the inventory fileansible -m setup wiki.example.com
Use the setup module to gather ansible 'facts' (aka ansible_variables) about that host.ansible localhost -m setup -a 'gather_subset=!all'
or look at the localhostansible all -m setup -a "filter=ansible_distribution*"
use a filter action to see specific variablesansible localhost -m setup --tree /tmp/facts
store all facts in a file 'tree', based on hostnameansible -m debug -a "var=hostvars['wiki.example.com']" localhost
gives you the 'ansible hostvars'ansible-playbook play1.yml play2.yml
Run multiple playbooksansible-playbook -i production webservers.yml --tags ntp --list-tasks
confirm what task names would be run if I ran this command and said "just ntp tasks"ansible-playbook --list-tags launch.yml
see what tags exist in my playbook (the tasks list shows more detail + the tags)ansible-playbook -i production webservers.yml --limit boston --list-hosts
confirm what hostnames might be communicated with if I said "limit to boston" [3]~/bin/ansible/contrib/inventory/digital_ocean.py --list --pretty --api-token TOKEN_HERE
use the DO api to list your droplets (dynamic inventory)ansible-playbook -vvv launch.yml -l wiki.example.com --user=root -e do_name=wiki.example.com --start-at-task='remove empty wiki schema from database if it already exists'
start at a particular point in the task listphp -r 'var_dump(json_decode(file_get_contents("/tmp/facts/localhost"), true));'
look at the json with php (or more interesting tools) With Ansible's Jinja2 filters, you can specify the output of a variable to be 'pretty' {{ some_variable | to_nice_json }}
Variables
- You have 3 plays in one playbook. Will play 3 be able to reference facts registered in play 1?
- facts, yes, play vars, no
- vars associated to the host, persist, vars defined in the play, do not, set_facts, registered vars and gathered facts associate to the host so those do persist for the run
Playbooks
Ansible "Playbooks" use an easy and descriptive language based on YAML.
Targets
Ansible can deploy to virtualization environments and public and private cloud environments including VMWare, OpenStack, AWS, Eucalyptus Cloud, KVM, and CloudStack
Testing
Jeff Geerling talks about the spectrum of testing you can employ in your Ansible deployments[4]
yamllint
ansible-playbook --syntax-check
ansible-lint
- molecule test (integration)
ansible-playbook --check
(against prod)- parallel infrastructure
In development
- Use the debug module
- Use the fail module to fail
- Use the assert module to make assertions (and fail if they don't match)
Best Practices
Building Ansible Automation Platform execution environments (EE)
Using Python 3
- https://docs.ansible.com/ansible/latest/reference_appendices/python_3_support.html
- https://docs.ansible.com/ansible/latest/dev_guide/developing_python_3.html
- Use tags to organize your Ansible work
- Use caching (default is off) to be able to refer to host 'facts' without having to hit each host in a playbook.
- Use register of variables to create more 'facts'. Results vary from module to module. Use -v to see possible values.
- There is an order of precedence with playbook variables, with role defaults the lowest priority and extra vars the winner.
- The array notation is preferred over the dot notation for accessing variables.
{{ ansible_eth0["ipv4"]["address"] }} over {{ ansible_eth0.ipv4.address }} because some keywords in Python would conflict
- Reserved words:
- hostvars
- group_names
- groups
- environemnt
inventory_hostname
is the name of the hostname as configured in Ansible's inventory host file.ansible_hostname
is the discovered hostname
- You can use a variables file to put sensitive data in a different file (one excluded from git).
- hosts: all
remote_user: root
vars:
favcolor: blue
vars_files:
- /vars/top_secret.yml
- You can use variables on the command line (and besides key=value pairs, you can use json or yml)
---
- hosts: '{{ hosts }}'
remote_user: '{{ user }}'
tasks:
- ...
ansible-playbook release.yml --extra-vars "hosts=vipers user=starbuck"
- Check Performance Tuning like enabling
pipelining
which is off by default.
- Truthy values should always be expressed as one of
[false, true]
. Although the Ansible docs show that you can use several forms of expression for boolean values, and the YAML spec specifies a fuller range of possibilities described below, the Ansible documentation now also clarifies that only lowercase 'true' or 'false' is compatible with yamllint options.
- YAML boolean values, not Ansible
y|Y|yes|Yes|YES|n|N|no|No|NO |true|True|TRUE|false|False|FALSE |on|On|ON|off|Off|OFF
Scope
Ansible has 3 main scopes:
Global: this is set by config, environment variables and the command line Play: each play and contained structures, vars entries, include_vars, role defaults and vars. Host: variables directly associated to a host, like inventory, facts or registered task outputs
Ansible with Vagrant
https://docs.ansible.com/ansible/guide_vagrant.html and Private:QualityBox/Vagrant
Ansible with MediaWiki
https://github.com/Orain I've cloned the 'ansible-playbook'
Ansible with Drupal
- Jeff Geerling (geerlingguy) has his code on github https://github.com/geerlingguy/drupal-vm, and also a website http://www.drupalvm.com/. He's the author of Ansible for DevOps. The only problem I see with his code is that it installs everything from his own 'roles' (individual components) via the sharing site/mechanism called Ansible Galaxy. So, for example, phpMyAdmin comes from https://github.com/geerlingguy/ansible-role-phpmyadmin This is good in that he can make his system work, but it's bad in that you're getting all your bits from him and can't tweak any of it without manually checking each role for the code and instructions behind it so you know what you can set via variables and such. I'd rather see each of these roles contained in the project, community sourced, installed via git.
geerlingguy.firewall geerlingguy.git geerlingguy.apache geerlingguy.memcached geerlingguy.mysql geerlingguy.php geerlingguy.php-pecl geerlingguy.php-memcached geerlingguy.php-mysql geerlingguy.php-xdebug geerlingguy.php-xhprof geerlingguy.phpmyadmin geerlingguy.composer geerlingguy.drush geerlingguy.daemonize geerlingguy.mailhog geerlingguy.java geerlingguy.solr
Ansible in the cloud
Ansible has several core modules for working with various cloud providers. These include
- Amazon
- Digital Ocean http://docs.ansible.com/digital_ocean_module.html
- Linode http://docs.ansible.com/linode_module.html
- LXC
- OpenStack
Ansible on Fedora
The Fedora Project uses Ansible in it's Infrastructure team, and they publish their whole setup https://infrastructure.fedoraproject.org/cgit/ansible.git/tree/README
Ansible Docs
Some of the docs pages I've visited
- https://docs.ansible.com/ansible/playbooks_intro.html
- https://docs.ansible.com/ansible/intro_inventory.html
- http://docs.ansible.com/playbooks_best_practices.html
- http://docs.ansible.com/playbooks_loops.html
- https://docs.ansible.com/ansible/playbooks_conditionals.html
- https://docs.ansible.com/ansible/playbooks_startnstep.html
- https://docs.ansible.com/ansible/playbooks_roles.html#task-include-files-and-encouraging-reuse
- http://docs.ansible.com/YAMLSyntax.html
- https://docs.ansible.com/ansible/become.html
- https://docs.ansible.com/ansible/debug_module.html
- https://docs.ansible.com/ansible/playbooks_debugger.html (
strategy:debug
) - https://docs.ansible.com/ansible/playbooks_conditionals.html#sts=The When Statement%C2%B6
Ansible References
- http://tjelvarolsson.com/blog/taking-the-effort-out-of-server-configuration-using-ansible/
- http://tjelvarolsson.com/blog/how-to-create-automated-and-reproducible-work-flows-for-installing-scientific-software/ < with Vagrant
- http://jpmens.net/2012/06/06/configuration-management-with-ansible/
- Jinja - the template engine for Ansible
References
- ↑ https://stackoverflow.com/questions/25230376/how-to-automatically-install-ansible-galaxy-roles
- ↑ Supposedly this only works for newer versions of Ansible, per the warning on their homepage:
Warning alert:To be able to download content from galaxy it is required to have ansible-core>=2.13.9 Please, check it running the command: ansible --version
But, it worked fine for me in the Meza 1_39 upgrade using Ansible 2.9.27
- ↑ Choosing which host(s) to operate on https://docs.ansible.com/ansible/intro_patterns.html
- ↑ https://www.youtube.com/live/FaXVZ60o8L8?si=gFoxE-ig5X0psuul&t=1244