In a playbook, you may want to execute different tasks, or have different goals, depending on the value of a fact (data about the remote system), a variable, or the result of a previous task. You may want the value of some variables to depend on the value of other variables. Or you may want to create additional groups of hosts based on whether the hosts match other criteria. You can do all of these things with conditionals.
Ansible uses Jinja2 :ref:`tests <playbooks_tests>` and :ref:`filters <playbooks_filters>` in conditionals. Ansible supports all the standard tests and filters, and adds some unique ones as well.
There are many options to control execution flow in Ansible. You can find more examples of supported conditionals at `<https://jinja.palletsprojects.com/en/master/templates/#comparisons>`_.
The simplest conditional statement applies to a single task. Create the task, then add a ``when`` statement that applies a test. The ``when`` clause is a raw Jinja2 expression without double curly braces (see :ref:`group_by_module`). When you run the task or playbook, Ansible evaluates the test for all hosts. On any host where the test passes (returns a value of True), Ansible runs that task. For example, if you are installing mysql on multiple machines, some of which have SELinux enabled, you might have a task to configure SELinux to allow mysql to run. You would only want that task to run on machines that have SELinux enabled:
..code-block:: yaml
tasks:
- name: Configure SELinux to start mysql on any port
Often you want to execute or skip a task based on facts. Facts are attributes of individual hosts, including IP address, operating system, the status of a filesystem, and many more. With conditionals based on facts:
See :ref:`commonly_used_facts` for a list of facts that frequently appear in conditional statements. Not all facts exist for all hosts. For example, the 'lsb_major_release' fact used in an example below only exists when the lsb_release package is installed on the target host. To see what facts are available on your systems, add a debug task to your playbook::
You can use `logical operators <https://jinja.palletsprojects.com/en/master/templates/#logic>`_ to combine conditions. When you have multiple conditions that all need to be true (that is, a logical ``and``), you can specify them as a list::
If a fact or variable is a string, and you need to run a mathematical comparison on it, use a filter to ensure that Ansible reads the value as an integer::
tasks:
- shell: echo "only on Red Hat 6, derivatives, and later"
when: ansible_facts['os_family'] == "RedHat" and ansible_facts['lsb']['major_release'] | int >= 6
.._conditionals_registered_vars:
Conditions based on registered variables
----------------------------------------
Often in a playbook you want to execute or skip a task based on the outcome of an earlier task. For example, you might want to configure a service after it is upgraded by an earlier task. To create a conditional based on a registered variable:
#. Register the outcome of the earlier task as a variable.
#. Create a conditional test based on the registered variable.
You create the name of the registered variable using the ``register`` keyword. A registered variable always contains the status of the task that created it as well as any output that task generated. You can use registered variables in templates and action lines as well as in conditional ``when`` statements. You can access the string contents of the registered variable using ``variable.stdout``. For example::
- name: test play
hosts: all
tasks:
- shell: cat /etc/motd
register: motd_contents
- shell: echo "motd contains the word hi"
when: motd_contents.stdout.find('hi') != -1
You can use registered results in the loop of a task if the variable is a list. If the variable is not a list, you can convert it into a list, with either ``stdout_lines`` or with ``variable.stdout.split()``. You can also split the lines by other fields::
- name: registered variable usage as a loop list
hosts: all
tasks:
- name: retrieve the list of home directories
command: ls /home
register: home_dirs
- name: add home dirs to the backup spooler
file:
path: /mnt/bkspool/{{ item }}
src: /home/{{ item }}
state: link
loop: "{{ home_dirs.stdout_lines }}"
# same as loop: "{{ home_dirs.stdout.split() }}"
The string content of a registered variable can be empty. If you want to run another task only on hosts where the stdout of your registered variable is empty, check the registered variable's string contents for emptiness:
..code-block:: yaml
- name: check registered variable for emptiness
hosts: all
tasks:
- name: list contents of directory
command: ls mydir
register: contents
- name: check contents for emptiness
debug:
msg: "Directory is empty"
when: contents.stdout == ""
Ansible always registers something in a registered variable for every host, even on hosts where a task fails or Ansible skips a task because a condition is not met. To run a follow-up task on these hosts, query the registered variable for ``is skipped`` (not for "undefined" or "default"). See :ref:`registered_variables` for more information. Here are sample conditionals based on the success or failure of a task. Remember to ignore errors if you want Ansible to continue executing on a host when a failure occurs:
..note:: Older versions of Ansible used ``success`` and ``fail``, but ``succeeded`` and ``failed`` use the correct tense. All of these options are now valid.
You can also create conditionals based on variables defined in the playbooks or inventory. Because conditionals require boolean input (a test must evaluate as True to trigger the condition), you must apply the ``| bool`` filter to non boolean variables, such as string variables with content like 'yes', 'on', '1', or 'true'. You can define variables like this:
If you combine a ``when`` statement with a :ref:`loop <playbooks_loops>`, Ansible processes the condition separately for each item. This is by design, so you can execute the task on some items in the loop and skip it on other items. For example:
If you need to skip the whole task when the loop variable is undefined, use the `|default` filter to provide an empty iterator. For example, when looping over a list:
You can provide your own facts, as described in :ref:`developing_modules`. To run them, just make a call to your own custom fact gathering module at the top of your list of tasks, and variables returned there will be accessible to future tasks:
You can use conditionals with re-usable tasks files, playbooks, or roles. Ansible executes these conditional statements differently for dynamic re-use (includes) and for static re-use (imports). See :ref:`playbooks_reuse` for more information on re-use in Ansible.
When you add a conditional to an import statement, Ansible applies the condition to all tasks within the imported file. This behavior is the equivalent of :ref:`tag_inheritance`. Ansible applies the condition to every task, and evaluates each task separately. For example, you might have a playbook called ``main.yml`` and a tasks file called ``other_tasks.yml``::
Thus if ``x`` is initially undefined, the ``debug`` task will be skipped. If this is not the behavior you want, use an ``include_*`` statement to apply a condition only to that statement itself.
You can apply conditions to ``import_playbook`` as well as to the other ``import_*`` statements. When you use this approach, Ansible returns a 'skipped' message for every task on every host that does not match the criteria, creating repetitive output. In many cases the :ref:`group_by module <group_by_module>` can be a more streamlined way to accomplish the same objective; see :ref:`os_variance`.
When you use a conditional on an ``include_*`` statement, the condition is applied only to the include task itself and not to any other tasks within the included file(s). To contrast with the example used for conditionals on imports above, look at the same playbook and tasks file, but using an include instead of an import::
By using ``include_tasks`` instead of ``import_tasks``, both tasks from ``other_tasks.yml`` will be executed as expected. For more information on the differences between ``include`` v ``import`` see :ref:`playbooks_reuse`.
- Add the same condition or conditions to all tasks in the role by placing your ``when`` statement under the ``roles`` keyword. See the example in this section.
- Add the same condition or conditions to all tasks in the role by placing your ``when`` statement on a static ``import_role`` in your playbook.
- Add a condition or conditions to individual tasks or blocks within the role itself. This is the only approach that allows you to select or skip some tasks within the role based on your ``when`` statement. To select or skip tasks within the role, you must have conditions set on individual tasks or blocks, use the dynamic ``include_role`` in your playbook, and add the condition or conditions to the include. When you use this approach, Ansible applies the condition to the include itself plus any tasks in the role that also have that ``when`` statement.
When you incorporate a role in your playbook statically with the ``roles`` keyword, Ansible adds the conditions you define to all the tasks in the role. For example:
Sometimes the facts about a host determine the values you want to use for certain variables or even the file or template you want to select for that host. For example, the names of packages are different on CentOS and on Debian. The configuration files for common services are also different on different OS flavors and versions. To load different variables file, templates, or other files based on a fact about the hosts you are managing:
Ansible separates variables from tasks, keeping your playbooks from turning into arbitrary code with nested conditionals. This approach results in more streamlined and auditable configuration rules because there are fewer decision points to track.
You can create a playbook that works on multiple platforms and OS versions with a minimum of syntax by placing your variable values in vars files and conditionally importing them. If you want to install Apache on some CentOS and some Debian servers, create variables files with YAML keys and values. For example::
Ansible gathers facts on the hosts in the webservers group, then interpolates the variable "ansible_facts['os_family']" into a list of filenames. If you have hosts with Red Hat operating systems ('CentOS', for example), Ansible looks for 'vars/RedHat.yml'. If that file does not exist, Ansible attempts to load 'vars/os_defaults.yml'. For Debian hosts, Ansible first looks for 'vars/Debian.yml', before falling back on 'vars/os_defaults.yml'. If no files in the list are found, Ansible raises an error.
You can use the same approach when different OS flavors or versions require different configuration files or templates. Select the appropriate file or template based on the variables assigned to each host. This approach is often much cleaner than putting a lot of conditionals into a single template to cover multiple OS or package versions.
