I started with kdump. It is pretty straightforward - clean yaml, no python code.
Next I tried storage - not trivial - several python and ansible issues.
I haven't tried any of the other ones.
If we convert all of the roles to use tox, should we ignore the CI failures, then go back later and file PRs to fix the issues? Or should "convert role to use tox" mean "convert role to use tox and CI should pass 100%"?
systemroles@lists.fedorahosted.org