async status error when used with become...

[Julien Tognazzi]

Hi all,

I notice a weird behavior regarding async_status

Assuming user1 and user2 exists on the remote host (here localhost)

async.yml

---

- hosts: localhost

  become: true

  tasks:

  - name: simulate long running op, allow to run for 45 sec, fire and forget

    command: /bin/sleep 15

    become_user: user2

    async: 45

    poll: 0

    register: async_task

  - name: check async job in same play

    async_status: jid={{ async_task.ansible_job_id }}

    become_user: user2

- hosts: localhost

  tasks:

  - name: check async job in another play

    async_status: jid={{ async_task.ansible_job_id }}

    become_user: user2

The result is:

user1@localhost ~$ ansible-playbook test/async.yml -vv

ansible-playbook 2.4.0.0

  config file = /home/user1/ansible.cfg

  configured module search path = [u'/home/user1/.ansible/plugins/modules', u'/usr/share/ansible/plugins/modules']

  ansible python module location = /usr/lib/python2.7/site-packages/ansible

  executable location = /usr/bin/ansible-playbook

  python version = 2.7.5 (default, Aug  4 2017, 00:39:18) [GCC 4.8.5 20150623 (Red Hat 4.8.5-16)]

Using /home/user1/ansible.cfg as config file

 [WARNING]: Could not match supplied host pattern, ignoring: all

 [WARNING]: provided hosts list is empty, only localhost is available

PLAYBOOK: async.yml ****************************************************************************************************

2 plays in test/async.yml

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

TASK [Gathering Facts] *************************************************************************************************

ok: [localhost]

META: ran handlers

TASK [simulate long running op, allow to run for 45 sec, fire and forget] **********************************************

task path: /home/user1/test/async.yml:7

changed: [localhost] => {"ansible_job_id": "182826547249.21334", "changed": true, "failed": false, "finished": 0, "results_file": "/opt/user2/.ansible_async/182826547249.21334", "started": 1}

TASK [check async job in same play] ************************************************************************************

task path: /home/user1/test/async.yml:14

ok: [localhost] => {"ansible_job_id": "182826547249.21334", "changed": false, "failed": false, "finished": 0, "started": 1}

META: ran handlers

META: ran handlers

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

TASK [Gathering Facts] *************************************************************************************************

ok: [localhost]

META: ran handlers

TASK [check async job in another play] *********************************************************************************

task path: /home/user1/test/async.yml:21

fatal: [localhost]: FAILED! => {"ansible_job_id": "182826547249.21334", "changed": false, "failed": true, "finished": 1, "msg": "could not find job", "started": 1}

        to retry, use: --limit @/home/user1/test/async.retry

PLAY RECAP *************************************************************************************************************

localhost                  : ok=4    changed=1    unreachable=0    failed=1

here, /home/user1/ansible.cfg is just an empty file.

the first play we run as root then become user2

On the second play we run directly as user1 and become user2...

is this a bug regarding permissions or am I missing something ?

Regards,

Julien