I don't see why we need to move anything as long as we can produce the results and artifacts and have it be outward facing.
I am currently getting all atomic fedora, centos, and RHEL on our internal CI infrastructure.
The main thing missing for Centos that we have for Fedora and RHEL is a notification mechanism. We use fedmsg and CI message bus for those. I know Jan Chaloupke was advocating something similar for Centos.
I think it is a good idea since the mechanism for notification and triggering are essentially the same.
On Fri, 2015-06-12 at 15:36 -0400, Colin Walters wrote:
On Mon, Jun 8, 2015, at 04:24 PM, Karsten Wade wrote:
You would join the Atomic SIG and ask for credentials to build and run CI jobs for all of the above. In fact, you are probably welcome to move all your jobs to ci.centos.org as your single place to work from rather than the internal-to-RH CI.
I don't think it's realistic to have CI/testing decoupled in infrastructure from the rel-eng/delivery - you really want tight integration to support things like "promotion" workflows where a build gets shipped in a broader channel after testing.
That argues for a model where Fedora has an independent Jenkins that's ideally similar in capabilities and model to what CentOS has. I think most of the work in Fedora currently is around Taskotron, which... it's not really worth going into the details too much, but I'm sure a Jenkins instance could be dedicated to this.
We could stand up something in projectatomic.io which gathers a summary of reporting but does not actually perform testing on its own, like what we're doing for links to images.
CentOS-devel mailing list CentOS-devel@centos.org http://lists.centos.org/mailman/listinfo/centos-devel