While running a github linter workflow on a new CentOS VM that I set up, I got the following error:
Run ansible-lint --parseable ./ansible
Traceback (most recent call last):
File "/usr/local/bin/ansible-lint", line 8, in <module>
sys.exit(_run_cli_entrypoint())
File "/usr/local/lib/python3.9/site-packages/ansiblelint/__main__.py", line 344, in _run_cli_entrypoint
sys.exit(main(sys.argv))
File "/usr/local/lib/python3.9/site-packages/ansiblelint/__main__.py", line 197, in main
initialize_options(argv[1:])
File "/usr/local/lib/python3.9/site-packages/ansiblelint/__main__.py", line 110, in initialize_options
options.cache_dir = get_cache_dir(options.project_dir)
File "/usr/local/lib/python3.9/site-packages/ansible_compat/prerun.py", line 13, in get_cache_dir
basename = project_dir.resolve().name.encode(encoding="utf-8")
AttributeError: 'str' object has no attribute 'resolve'
Error: Process completed with exit code 1.
I tried with the following package versions which all failed with the above error:
- python3: 3.9.16
- ansible: 8.0.0 or 6.6.0
- ansible-compat: 4.1.2 or 2.2.7
- ansible-core: 2.15.0 or 2.13.9
- ansible-lint: 6.17.0 or 6.10.2
Using pip, the following dependencies were also installed:
- flake8
- yamllint==1.28.0
- codespell
- comment_parser
Is there some linter dependency or something else that I am missing?
The step within the linter job is:
- name: 'Lint yaml - ansible, GH actions and GH workflows'
run: yamllint -f parsable ./
if: always()
2
Answers
In my environment, downgrading ansible_compat to v3.0.1 with pip install solved the problem. From v4 on, prerun.py contains the line that gives error.
Seems to be an ansible-lint bug:
Latest ansible-lint release (v6.15.0) incompatible with latest ansible-compat (v4.0.2) · Issue #3408 · ansible/ansible-lint · GitHub
According to the upstream developer: