issue when using tox-docker on bitbucket pipelines since 20.0.0

[Ian Fiddes]

Hello,

My deployments stopped working today. I have traced it to something that changed between 20.0.0, and the pre-releases. It seems to not recognize the docker container, despite it being present:

+ tox -vv
using tox.ini: /opt/atlassian/pipelines/agent/build/tox.ini (pid 613)
Traceback (most recent call last):
  File "/opt/conda/bin/tox", line 8, in <module>
    sys.exit(cmdline())
  File "/opt/conda/lib/python3.7/site-packages/tox/session/__init__.py", line 44, in cmdline
    main(args)
  File "/opt/conda/lib/python3.7/site-packages/tox/session/__init__.py", line 65, in main
    config = load_config(args)
  File "/opt/conda/lib/python3.7/site-packages/tox/session/__init__.py", line 81, in load_config
    config = parseconfig(args)
  File "/opt/conda/lib/python3.7/site-packages/tox/config/__init__.py", line 283, in parseconfig
    pm.hook.tox_configure(config=config)  # post process config object
  File "/opt/conda/lib/python3.7/site-packages/pluggy/hooks.py", line 286, in __call__
    return self._hookexec(self, self.get_hookimpls(), kwargs)
  File "/opt/conda/lib/python3.7/site-packages/pluggy/manager.py", line 93, in _hookexec
    return self._inner_hookexec(hook, methods, kwargs)
  File "/opt/conda/lib/python3.7/site-packages/pluggy/manager.py", line 87, in <lambda>
    firstresult=hook.spec.opts.get("firstresult") if hook.spec else False,
  File "/opt/conda/lib/python3.7/site-packages/pluggy/callers.py", line 208, in _multicall
    return outcome.get_result()
  File "/opt/conda/lib/python3.7/site-packages/pluggy/callers.py", line 80, in get_result
    raise ex[1].with_traceback(ex[2])
  File "/opt/conda/lib/python3.7/site-packages/pluggy/callers.py", line 187, in _multicall
    res = hook_impl.function(*args)
  File "/opt/conda/lib/python3.7/site-packages/tox_docker/__init__.py", line 108, in tox_configure
    raise ValueError(f"{container_name!r} is not a valid container name")
ValueError: '####' is not a valid container name

[Dan Crosta]

Hm, can you show the [docker:...] sections from your tox.ini? This error implies you have something like `[docker:####]'

What was the last version that worked? It looks like the lines causing that error have existed since 2.0.0a1, so I'm surprised if the 2.0 final release caused an issue

[Ian Fiddes]

The docker section looks like this:

[docker:mycontainer:test]
ports = 5000:5000/tcp

Which is then included in the test sections:

[testenv]
docker =
    mycontainer:test

In regards to versions, I would need to do further testing. The error started happening yesterday, and the last successful build before the error was on January 26th. I had not been pegging the version of tox-docker, so this would be at the whim of whatever tox decided to install from pip. Looking at pypi now, it seems like the pre-release versions were tagged as such, and I believe that means that pip will ignore them unless asked specifically to install them.

[Dan Crosta]

Hm, OK. I assume test is a tag on the image?

In 2.0, the docker= directive in each testenv just gets a name that refers to another section; within the [docker:mycontainer] section, you can specify the image as image = mycontainer:test (or several other variations -- basically anything that docker pull would work with). This is a change from the 1.x series, and is one of the major reasons we cut a 2.x release.

If you pin your version of tox-docker to < 2.0, it probably will work, although you'll lose access to new features, like volumes/mounts and healthcheck support.

[Ian Fiddes]

Got it, thanks