[libvirt PATCH 0/4] ci: Drop Fedora 34, add Fedora 36

[Erik Skultety]

On Thu, May 26, 2022 at 05:15:28AM -0700, Andrea Bolognani wrote:
> On Thu, May 26, 2022 at 12:05:23PM +0200, Erik Skultety wrote:
> > On Thu, May 26, 2022 at 11:05:30AM +0200, Andrea Bolognani wrote:
> > > Test pipeline:
> > >
> > >   https://gitlab.com/abologna/libvirt/-/pipelines/548549670
> > >
> > > Matching libvirt-perl MR, needed because the integration tests
> > > download artifacts from there:
> > >
> > >   https://gitlab.com/libvirt/libvirt-perl/-/merge_requests/63
> >
> > Reviewed-by: Erik Skultety <eskultet at redhat.com>
> >
> > Please don't push this yet, because the integration tests would fail (as you
> > didn't didn't run them in your pipeline)
> 
> I didn't think I would be able to. Can anyone trigger a run of the
> integration tests? Can group members?

Any member can trigger the scheduled pipeline in the main repo, but you'd have
to set up your own runner for your fork to do it.

> 
> > I need to create a template
> > image for fedora-36 first on the baremetal host. I'll ping you when it's done.
> 
> I think we will start seeing failures in the Fedora 34 integration
> jobs now that the libvirt-perl MR has been accepted and thus the
> corresponding job is gone. Or is GitLab CI smart enough to go and
> pick the artifacts from the last existing occurrence of the job?
> 
> I can split things so that Fedora 34 is dropped and Fedora 36 is only
> added to the non-integration part of the pipeline at first, and then
> the integration job is added in a separate commit. The bulk could be
> pushed right away, with the last bit waiting for you to prepare the
> VM template.
> 
> Or is that a quick enough job that it doesn't matter in practice and
> I'm massively overthinking this? O:-)

Yeah, usually I can do it within minutes, but the pipeline failed [1]. I think
in the case of F35 there may be one more problem with your patches and for F36,
sigh, the failure relates to a recent CVE fix in git [2] which e.g. the Avocado
project already had to deal with.
So I agree, please strip the fedora 36 changes to the integration stage and you
can push the rest while I figure out how to deal with the failures.

[1] https://gitlab.com/eskultety/libvirt/-/pipelines/548707692
[2] https://github.blog/2022-04-12-git-security-vulnerability-announced/

Andrea, I take it you'll repost your series without the F36 integration changes
so that I can ACK separately, right?

Erik