[ansible] Ansible: Store command's stdout in new variable?

Inside my playbook I'd like to create a variable holding the output of an external command. Afterwards I want to make use of that variable in a couple of templates.

Here are the relevant parts of the playbook:

  tasks:
    - name: Create variable from command
      command: "echo Hello"
      register: command_output
    - debug: msg="{{command_output.stdout}}"

    - name: Copy test service
      template: src=../templates/test.service.j2 dest=/tmp/test.service
    - name: Enable test service
      shell: systemctl enable /tmp/test.service
    - name: Start test service
      shell: systemctl start test.service

and let's say this is my template:

[Unit]
Description=MyApp
After=docker.service
Requires=docker.service

[Service]
TimeoutStartSec=0
ExecStartPre=-/usr/bin/docker kill busybox1
ExecStartPre=-/usr/bin/docker rm busybox1
ExecStartPre=/usr/bin/docker pull busybox
ExecStart=/usr/bin/docker run --name busybox1 busybox /bin/sh -c "while true; do echo {{ string_to_echo }}; sleep 1; done"

[Install]
WantedBy=multi-user.target

(Notice the {{ string_to_echo }})

So what I'm basically looking for is a way to store the contents of command_output.stdout (which is generated/retrieved during the first task) in a new variable string_to_echo.
That variable I'd like to use in multiple templates afterwards.

I guess I could just use {{command_output.stdout}} in my templates, but I want to get rid of that .stdout for readability.

This question is related to ansible ansible-playbook

The answer is


You have to store the content as a fact:

- set_fact:
    string_to_echo: "{{ command_output.stdout }}"

If you want to go further and extract the exact information you want from the Playbook results, use JSON query language like jmespath, an example:

  - name: Sample Playbook
    // Fill up your task
    no_log: True
    register: example_output

  - name: Json Query
    set_fact:
      query_result:
        example_output:"{{ example_output | json_query('results[*].name') }}"

A slight modification beyond @udondan's answer. I like to reuse the registered variable names with the set_fact to help keep the clutter to a minimum.

So if I were to register using the variable, psk, I'd use that same variable name with creating the set_fact.

Example

- name: generate PSK
  shell: openssl rand -base64 48
  register: psk
  delegate_to: 127.0.0.1
  run_once: true

- set_fact: 
    psk={{ psk.stdout }}

- debug: var=psk
  run_once: true

Then when I run it:

$ ansible-playbook -i inventory setup_ipsec.yml

 PLAY                                                                                                                                                                                [all] *************************************************************************************************************************************************************************

 TASK [Gathering                                                                                                                                                                     Facts] *************************************************************************************************************************************************************
 ok: [hostc.mydom.com]
 ok: [hostb.mydom.com]
 ok: [hosta.mydom.com]

 TASK [libreswan : generate                                                                                                                                                          PSK] ****************************************************************************************************************************************************
 changed: [hosta.mydom.com -> 127.0.0.1]

 TASK [libreswan :                                                                                                                                                                   set_fact] ********************************************************************************************************************************************************
 ok: [hosta.mydom.com]
 ok: [hostb.mydom.com]
 ok: [hostc.mydom.com]

 TASK [libreswan :                                                                                                                                                                   debug] ***********************************************************************************************************************************************************
 ok: [hosta.mydom.com] => {
     "psk": "6Tx/4CPBa1xmQ9A6yKi7ifONgoYAXfbo50WXPc1kGcird7u/pVso/vQtz+WdBIvo"
 }

 PLAY                                                                                                                                                                                RECAP *************************************************************************************************************************************************************************
 hosta.mydom.com    : ok=4    changed=1    unreachable=0    failed=0
 hostb.mydom.com    : ok=2    changed=0    unreachable=0    failed=0
 hostc.mydom.com    : ok=2    changed=0    unreachable=0    failed=0

In case than you want to store a complex command to compare text result, for example to compare the version of OS, maybe this can help you:

tasks:
       - shell: echo $(cat /etc/issue | awk {'print $7'})
         register: echo_content

       - shell: echo "It works"
         when: echo_content.stdout == "12"
         register: out
       - debug: var=out.stdout_lines

I'm a newbie in Ansible, but I would suggest next solution:

playbook.yml

...
vars:
  command_output_full:
    stdout: will be overriden below
  command_output: {{ command_output_full.stdout }}
...
...
...
tasks:
  - name: Create variable from command
    command: "echo Hello"
    register: command_output_full
  - debug: msg="{{ command_output }}"

It should work (and works for me) because Ansible uses lazy evaluation. But it seems it checks validity before the launch, so I have to define command_output_full.stdout in vars.

And, of course, if it is too many such vars in vars section, it will look ugly.


There's no need to set a fact.

    - shell: cat "hello"
      register: cat_contents

    - shell: echo "I cat hello"
      when: cat_contents.stdout == "hello"