fix(_common): replace deprecated vars dictionary access

[willbuckner]

Summary

• Replace direct membership checks against Ansible's internal vars dictionary with ansible.builtin.varnames lookup.

• Update all three deprecated variable assertions in the _common role.

• Preserve detection when deprecated variables are assigned empty values.

  Motivation

ansible-core 2.21 deprecated direct access to its internal vars dictionary, for removal in ansible-core 2.24. Currently I'm seeing these deprecation warnings while doing ansible deployments.

This is an alternative to #793 which is stale since April and failing some CI tests.

Using varnames avoids an undefined variable error when a variable is absent, and checks variable presence independently of its value.

Fixes #853

Validation

• ansible-lint and yamllint passed with zero failures or warnings.
• Verified with ansible-core 2.21.2 that an absent variable passes the assertion, and that an explicitly empty deprecated variable fails it.
• git diff --check passed.

[willbuckner]

Looks like a new set of integration tests ran and I have some failures--investigating.

Edit: Hmm, I'm not sure if my PR is causing this failure or not:

> WARNING: apt does not have a stable CLI interface. Use with caution in scripts.
  
  Hit:1 http://security.ubuntu.com/ubuntu focal-security InRelease
  Hit:2 http://archive.ubuntu.com/ubuntu focal InRelease
  Hit:3 http://archive.ubuntu.com/ubuntu focal-updates InRelease
  Hit:4 http://ppa.launchpad.net/deadsnakes/ppa/ubuntu focal InRelease
  Hit:5 http://archive.ubuntu.com/ubuntu focal-backports InRelease
  Reading package lists...
  Building dependency tree...
  Reading state information...
  108 packages can be upgraded. Run 'apt list --upgradable' to see them.
  
  WARNING: apt does not have a stable CLI interface. Use with caution in scripts.
  
  Reading package lists...
  Building dependency tree...
  Reading state information...
  docker.io is already the newest version (26.1.3-0ubuntu1~20.04.1).
  0 upgraded, 0 newly installed, 0 to remove and 108 not upgraded.
  ERROR    Failed to validate /root/ansible_collections/prometheus/prometheus/roles/bind_exporter/molecule/alternative/molecule.yml
  
  
  Command exited with status 1 after 2.9715819358825684 seconds.
  ERROR: Command "./runme.sh -vvvvvv" returned exit status 1.
  FATAL: The 1 integration test(s) listed below (out of 1) failed. See error output above for details:
  molecule-bind_exporter-alternative
  Run command with stdout: docker exec -i ansible-test-controller-Yj3XHefU sh -c 'tar cf - -C /root/ansible_collections/prometheus/prometheus/tests --exclude .tmp output | gzip'
  Run command with stdin: tar oxzf - -C /home/runner/work/ansible/ansible/ansible_collections/prometheus/prometheus/tests
  FATAL: Command "docker exec ansible-test-controller-Yj3XHefU /usr/bin/env ANSIBLE_TEST_CONTENT_ROOT=/root/ansible_collections/prometheus/prometheus LC_ALL=en_US.UTF-8 /usr/bin/python3.12 /root/ansible/bin/ansible-test integration -v --allow-destructive --containers '{}' --retry-on-error --continue-on-error --diff molecule-bind_exporter-alternative --truncate 0 --color yes --host-path tests/output/.tmp/host-xa1igg4u --metadata tests/output/.tmp/metadata-g58hoj0z.json" returned exit status 1.

[willbuckner]

I had Codex help me debug the CI failures a bit, and it seems unrelated to my PRs, but in case this helps track it down:

The actual hidden validation error is:

Additional properties are not allowed ('cgroup_parent' was
unexpected)

What happened:

• CI installs molecule and molecule-plugins[docker] without version
  constraints in tests/integration/requirements.txt:3.

• It selected Molecule 26.6.0 and molecule-plugins 26.7.15.

• molecule-plugins 26.7.15, released July 15, introduced a strict
  Docker driver schema. (release

  (https://github.com/ansible-community/molecule-plugins/releases/tag/v26.7.15),
  schema PR
  (feat(docker): Add driver schema ansible-community/molecule-plugins#340))

• The repository’s shared .config/molecule/config.yml:15 supplies
  cgroup_parent: docker.slice, which that schema rejects.

• Because _common changed, the label workflow intentionally enables
  tests for every role (.github/workflows/conventional-
  label.yml:99). Hence the enormous failure matrix.

I reproduced the error locally using CI’s exact dependency
versions. I also found the same failure and versions in an
unrelated July 27 run
(https://github.com/prometheus-community/ansible/actions/runs/30250652077),
before PR #869.

At my final check, all 548 failures were Molecule integration jobs;
there were zero failures outside that matrix.

The cleanest likely fix is a separate CI-compatibility change
removing cgroup_parent, because the current Docker plugin neither
accepts nor forwards it. A temporary alternative is pinning
molecule-plugins<26.7.15 while pursuing upstream support.

I don't know enough about molecule nor the rest of your CI system to really propose a fix for the CI failures to be honest.

[SuperQ]

It seems like one of the config items in our molecule config (cgroup_parent) is not allowed in the new schema test. Will see if I can fix this.

[SuperQ]

#871 should fix the tests.

[willbuckner]

@SuperQ CI passed this time after your change, thanks for doing that! This should be good to merge whenever.

[github-actions]

Docs Build 📝

Thank you for contributing!✨

This PR has been merged and the docs are now incorporated into main:
https://prometheus-community.github.io/ansible/branch/main