AnsibleRunnerV2 broken on upstream changes to PlaybookCLI in devel

[jctanner]

(venv) [jtanner@jtw530 role.1]$ python -c 'from testinfra.utils.ansible_runner import AnsibleRunnerV2; runner = AnsibleRunnerV2()'
Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/data/workspace.issues/AP-MOLECULE_COLLECTION_INIT/venv/lib/python3.7/site-packages/testinfra-1.16.0-py3.7.egg/testinfra/utils/ansible_runner.py", line 153, in __init__
    self.cli.options = self.cli.base_parser(
AttributeError: 'PlaybookCLI' object has no attribute 'base_parser'
(venv) [jtanner@jtw530 role.1]$

Caused by ansible/ansible@7e92ff8#diff-b54c875fc6bfd4d98dcc488f06a07bea

[singuliere]

The commit introducing the change, from the PR and before the squash post merge: ansible/ansible@39e4421

[cboylan]

Using https://github.com/ansible/ansible-runner may be more reliable long term method of running ansible from testinfra. I don't have enough knowledge around testinfra or ansible-runner to know for sure, but wanted to make sure this alternative was called out since ansible-runner seems built for this exact use case.

[jctanner]

I've already started a port to runner and it's only held up right now by an obscure bug: ansible/ansible-runner#200

[ianw]

@jctanner thanks for that; I took that gist and got it working for at least a loopback situation and will test more. I missing what was happening with the event capture, ascii codes and molecule, but I hope that with you as the author you can also update the pull request (or make new ones, whatever :) OpenStack should be able to test that pull request as it progresses in our CI too, as an extra level of testing over travis

#410

[philpep]

Great ! +1 for using ansible-runner

[decentral1se]

🚀 🚀 🚀

ansible-runner integration would indeed be fantastic.