Looping or repeating a group of tasks until success

Question

I currently have a playbook which includes a task file. In that task file, I would like to check for a condition. If the exit code of that condition is not equal to 0, all steps in the task file should be repeated. I have tried a few variations with block and loops but I have not figured out a way to make it do what I described above.

Currently I have something like this:

tasks:
  - name: call task file
    include: task_file.yml

In task_file.yml,

- name: perform an operations
  shell: do A
    
- name: check
  shell: do B
  register: result

Next, I would like to tell the main playbook that if result.rc != 0, please repeat task_file.yml until result.rc == 0.

Any pointers would be greatly appreciated

The playbook seems to end no matter what the exit code.

Answer 1

There's no direct way to reach your goal as include_tasks does not support the retry/until loop keywords.

There is an attempt to circumvent that limitation by teaching ansible a new loop_control.until keyword for loops which could be used for includes. Unfortunately, the pull request has been opened since Sep. 2019 and has still not reached a realease.

The good news is you can implement that with some work by using include recursion with a block. The below example is largely inspired by a blog article on https://dev.to. I adapted to the current context, fixed some good practice and added features like flexible retries number and delay between retries. Here we go:

The tasks to be retried go in task_file.yml

---
- name: group of tasks to repeat until success
  block:
    - name: increment attempts counter
      ansible.builtin.set_fact:
        attempt_number: "{{ attempt_number | d(0) | int + 1 }}"
        
    - name: dummy task
      ansible.builtin.debug:
        msg: "I'm a dummy task"

    - name: task to check for success.
      # Just for the example. Will return success on attempt number 3
      ansible.builtin.command: "[ {{ attempt_number | int }} -eq 3 ]"
      changed_when: false

  rescue:
    - name: "Fail if we reached the max of {{ max_attempts | d(3) }} attempts"
      # Default will be 3 attempts if max_attempts is not passed as a parameter
      ansible.builtin.fail:
        msg: Maximum number of attempts reached
      when: attempt_number | int == max_attempts | int | d(3)

    - ansible.builtin.debug:
        msg: "group of tasks failed on attempt {{ attempt_number }}. Retrying"

    - name: add delay if needed
      # no delay if retry_delay is not passed as parameter
      ansible.builtin.wait_for:
        timeout: "{{ retry_delay | int | d(omit) }}"
      when: retry_delay is defined

    # include ourselves to retry.
    - ansible.builtin.include_tasks: task_file.yml

As you can see, the file includes itself again in case of failure until success of max attempts is reached. Also, note that a retry will happen if any task fails inside the block, not only the last one. If you have a more complex scenario, you can implement more checks for fail/not fail in the rescue section an even add an always section if needed. See anbile blocks

Then you can call this file from your playbook:

---
- hosts: localhost
  gather_facts: false

  tasks:
    - name: Include tasks to retry. 7 attempts max with 1 second delay
      ansible.builtin.include_tasks: task_file.yml
      vars:
        max_attempts: 7
        retry_delay: 1

Playing this example succeeds on third attempt as hardcoded and expected. (You can play around with the parameters to test a fail scenario)

$ ansible-playbook playbook.yml

PLAY [localhost] ***********************************************************************************************************************************************************************************************************************

TASK [Include tasks to retry] **********************************************************************************************************************************************************************************************************
included: /tmp/toto/task_file.yml for localhost

TASK [increment attempts counter] ******************************************************************************************************************************************************************************************************
ok: [localhost]

TASK [dummy task] **********************************************************************************************************************************************************************************************************************
ok: [localhost] => {
    "msg": "I'm a dummy task"
}

TASK [task to check for success.] *******************************************************************************************************************************************************************************************************
fatal: [localhost]: FAILED! => {"changed": false, "cmd": ["[", "1", "-eq", "3", "]"], "delta": "0:00:00.002104", "end": "2022-12-08 14:16:27.850578", "msg": "non-zero return code", "rc": 1, "start": "2022-12-08 14:16:27.848474", "stderr": "", "stderr_lines": [], "stdout": "", "stdout_lines": []}

TASK [Fail if we reached the max of 7 attempts] ****************************************************************************************************************************************************************************************
skipping: [localhost]

TASK [ansible.builtin.debug] ***********************************************************************************************************************************************************************************************************
ok: [localhost] => {
    "msg": "group of tasks failed on attempt 1. Retrying"
}

TASK [add delay if needed] *************************************************************************************************************************************************************************************************************
ok: [localhost]

TASK [ansible.builtin.include_tasks] ***************************************************************************************************************************************************************************************************
included: /tmp/toto/task_file.yml for localhost

TASK [increment attempts counter] ******************************************************************************************************************************************************************************************************
ok: [localhost]

TASK [dummy task] **********************************************************************************************************************************************************************************************************************
ok: [localhost] => {
    "msg": "I'm a dummy task"
}

TASK [task to check for success.] *******************************************************************************************************************************************************************************************************
fatal: [localhost]: FAILED! => {"changed": false, "cmd": ["[", "2", "-eq", "3", "]"], "delta": "0:00:00.004009", "end": "2022-12-08 14:16:29.496509", "msg": "non-zero return code", "rc": 1, "start": "2022-12-08 14:16:29.492500", "stderr": "", "stderr_lines": [], "stdout": "", "stdout_lines": []}

TASK [Fail if we reached the max of 7 attempts] ****************************************************************************************************************************************************************************************
skipping: [localhost]

TASK [ansible.builtin.debug] ***********************************************************************************************************************************************************************************************************
ok: [localhost] => {
    "msg": "group of tasks failed on attempt 2. Retrying"
}

TASK [add delay if needed] *************************************************************************************************************************************************************************************************************
ok: [localhost]

TASK [ansible.builtin.include_tasks] ***************************************************************************************************************************************************************************************************
included: /tmp/toto/task_file.yml for localhost

TASK [increment attempts counter] ******************************************************************************************************************************************************************************************************
ok: [localhost]

TASK [dummy task] **********************************************************************************************************************************************************************************************************************
ok: [localhost] => {
    "msg": "I'm a dummy task"
}

TASK [task to check for success.] *******************************************************************************************************************************************************************************************************
ok: [localhost]

PLAY RECAP *****************************************************************************************************************************************************************************************************************************
localhost                  : ok=14   changed=0    unreachable=0    failed=0    skipped=2    rescued=2    ignored=0