At Summit, I talked briefly to Karanbir Singh from CentOS. CentOS has a
Jenkins system with a lot of hardware behind it (https://ci.centos.org/)
and they run all of their layered image builds* through that. KB
offered to run their tests on any images Fedora produces. That seems
cool, but I'm not sure how we could best take advantage of this. How
could we integrate such tests into our processes?
As a first step, does anyone have a problem with saying to CentOS,
"Cool! Start running those tests! Thanks!" — even if we don't know what
exactly what we're going to do with any results? I'm imagining that
once we get to $futurephase of the build service plan, layered images
will go through Bodhi just like RPMs, and at that point, this feedback
could be hooked in there. In the meantime, packagers of layered images
could simply be told to look manually.
The tests themselves, as I understand it, are documented at
<http://wiki.centos.org/QaWiki/AutomatedTests/WritingTests/t_functional>,
by the way, including contribution information.
* See change proposal for a layered image build service for Fedora —
actually testing images which are produced is a next step specifically
listed as out of scope for the change itself.
<https://fedoraproject.org/wiki/Changes/Layered_Docker_Image_Build_Service>
--
Matthew Miller
<mattdm(a)fedoraproject.org>
Fedora Project Leader