Closed justinc1 closed 1 year ago
Dry-running this example - https://github.com/ScaleComputing/HyperCoreAnsibleCollection/actions/runs/4687678208/jobs/8307275009#step:9:45. It does not complain about "'ansible_python_interpreter' is undefined" anymore.
It does fail later, because example is tested in check mode, so command is not really run, and we do not get stdout_lines. But this is not problem with code, it is problem how we test the examples.
https://github.com/ScaleComputing/HyperCoreAnsibleCollection/actions/runs/4686722374/jobs/8305135195#step:9:48 Here ansible_python_interpreter was not defined.
See also https://github.com/ansible/ansible/issues/21311 and https://docs.ansible.com/ansible/latest/reference_appendices/special_variables.html I missed before that ansible_python_interpreter is not always defined.
We do not want to force user to tell us some valid python interpreter. Also, it would be possible that the only availble python is in venv or in /usr/local/bin (python docker images).
ansible_playbook_python looks like a suitable fallback. It is interpreter on the controller node. Our script will always run there - we never ssh to HyperCore controller.