Working with Patterns

Patterns in Ansible are how we decide which hosts to manage. This can mean what hosts to communicate with, but in termsof Working With Playbooks it actually means what hosts to apply a particular configuration or IT process to.

We’ll go over how to use the command line in Introduction To Ad-Hoc Commands section, however, basically it looks like this:

  1. ansible <pattern_goes_here> -m <module_name> -a <arguments>

Such as:

  1. ansible webservers -m service -a "name=httpd state=restarted"

A pattern usually refers to a set of groups (which are sets of hosts) – in the above case, machines in the “webservers” group.

Anyway, to use Ansible, you’ll first need to know how to tell Ansible which hosts in your inventory to talk to.This is done by designating particular host names or groups of hosts.

The following patterns are equivalent and target all hosts in the inventory:

  1. all
  2. *

It is also possible to address a specific host or set of hosts by name:

  1. one.example.com
  2. one.example.com:two.example.com
  3. 192.0.2.50
  4. 192.0.2.*

The following patterns address one or more groups. Groups separated by a colon indicate an “OR” configuration.This means the host may be in either one group or the other:

  1. webservers
  2. webservers:dbservers

You can exclude groups as well, for instance, all machines must be in the group webservers but not in the group phoenix:

  1. webservers:!phoenix

You can also specify the intersection of two groups. This would mean the hosts must be in the group webservers andthe host must also be in the group staging:

  1. webservers:&staging

You can do combinations:

  1. webservers:dbservers:&staging:!phoenix

The above configuration means “all machines in the groups ‘webservers’ and ‘dbservers’ are to be managed if they are inthe group ‘staging’ also, but the machines are not to be managed if they are in the group ‘phoenix’ … whew!

You can also use variables if you want to pass some group specifiers via the “-e” argument to ansible-playbook, but thisis uncommonly used:

  1. webservers:!{{excluded}}:&{{required}}

You also don’t have to manage by strictly defined groups. Individual host names, IPs and groups, can also be referenced usingwildcards

  1. *.example.com
  2. *.com

It’s also ok to mix wildcard patterns and groups at the same time:

  1. one*.com:dbservers

You can select a host or subset of hosts from a group by their position. For example, given the following group:

  1. [webservers]
  2. cobweb
  3. webbing
  4. weber

You can refer to hosts within the group by adding a subscript to the group name:

  1. webservers[0]       # == cobweb
  2. webservers[-1]      # == weber
  3. webservers[0:2]     # == webservers[0],webservers[1]
  4.                     # == cobweb,webbing
  5. webservers[1:]      # == webbing,weber
  6. webservers[:3]      # == cobweb,webbing,weber

Most people don’t specify patterns as regular expressions, but you can. Just start the pattern with a ‘~’:

  1. ~(web|db).*\.example\.com

While we’re jumping a bit ahead, additionally, you can add an exclusion criteria just by supplying the —limit flag to /usr/bin/ansible or /usr/bin/ansible-playbook:

  1. ansible-playbook site.yml --limit datacenter2

And if you want to read the list of hosts from a file, prefix the file name with ‘@’.:

  1. ansible-playbook site.yml --limit @retry_hosts.txt

Easy enough. See Introduction To Ad-Hoc Commands and then Working With Playbooks for how to apply this knowledge.

Note

You can use ‘,’ instead of ‘:’ as a host list separator. The ‘,’ is preferred specially when dealing with ranges and ipv6.

See also