Hi, After some discussions, we came up with the idea to have the RDO project owns its build infrastructure. To play with this idea, we propose to experiment to use an OpenStack like workflow to build RDO packages by deploying a Software Factory instance (http://softwarefactory-project.io) on top of an RDO or RHEL-OSP cloud. That will allow us to have our own Gerrit, Zuul, Nodepool and Jenkins instances like in the OpenStack upstream project while adding our package building specific needs like using the Delorean and Mock/Koji/Mash machineries. The objectives from these changes are: 1. to have a full gating CI for RDO to never break the package repository. 2. to be in control of our infrastructure. 3. to simplify the work-flow where we can to make it more efficient and easier to grasp. Nothing is set in stone so feel free to comment or ask questions. Cheers, -- Fred - May the Source be with you _______________________________________________ Rdo-list mailing list Rdo-list(a)redhat.com https://www.redhat.com/mailman/listinfo/rdo-list To unsubscribe: rdo-list-unsubscribe(a)redhat.com

> As David said, our infrastructure is scattered and as been a > constantly moving target so I'd consider this a step toward the right > direction. Agreed but we also started consolidation around CentOS CloudSIG using CentOS provided infra and I think we should keep that direction. IMHO it would make sense to include this available hardware and this new RDO based cloud to ci.centos.org so we can continue already choosen consolidation direction. Cheers, Alan

As David said, our infrastructure is scattered and as been a constantly moving target so I'd consider this a step toward the right direction. The advantages of this proposal being: 1. stable infrastructure => less time spent on fixing the infrastructure 2. customizable workflow thanks to zuul which could only result in improving quality 3. close the gap between upstream and downstream infrastructure 4. self-hosted on RDO => that's a very important one I don't see any real negative points, so unless someone has a different proposal, I suggest that we start with a PoC. As the Liberty cycle is about to finish, we'll get some bandwidth to work on our infrastructure so this is the best time to discuss this. Consolidating our infrastructure is a primary goal to open up RDO governance further.

2015-10-14 11:30 GMT+02:00 Karanbir Singh <mail-lists(a)karan.org&gt;: > One thing that I dont understand here is what value this adds over the > CentOS Build Services.. we can integrate with an existing source control > setup ( or use git.c.o ) and we can do fairly extensive test hosting and > a release cadence built on that. > > Or is the intention here to host software-factory as a RDO specific UI > backed by the CentOS pipeline ? We would keep CBS for final, production builds. I'd like to imagine SF as an additional service provided in the CentOS community for all projects to use, not just RDO/Cloud SIG! Value add AFAICT is that it combines tools in one nice UI and enables automated workflows e.g. I would like to see bot proposing gerrit changes to bump versions in spec Requires: when upstream global-requirements are changed. I'm discovering what SF can do myself, but it looks promising. Cheers, Alan _______________________________________________ Rdo-list mailing list Rdo-list(a)redhat.com https://www.redhat.com/mailman/listinfo/rdo-list To unsubscribe: rdo-list-unsubscribe(a)redhat.com

On 10/14/2015 07:30 PM, Karanbir Singh wrote: > On 13/10/15 19:43, Haïkel wrote: >> As David said, our infrastructure is scattered and as been a >> constantly moving target so I'd consider this a step toward the right >> direction. >> >> The advantages of this proposal being: >> 1. stable infrastructure => less time spent on fixing the infrastructure >> 2. customizable workflow thanks to zuul which could only result in >> improving quality >> 3. close the gap between upstream and downstream infrastructure >> 4. self-hosted on RDO => that's a very important one >> >> I don't see any real negative points, so unless someone has a >> different proposal, I suggest that we start with a PoC. >> As the Liberty cycle is about to finish, we'll get some bandwidth to >> work on our infrastructure so this is the best time to discuss this. >> >> Consolidating our infrastructure is a primary goal to open up RDO >> governance further. >> > > One thing that I dont understand here is what value this adds over the > CentOS Build Services.. we can integrate with an existing source control > setup ( or use git.c.o ) and we can do fairly extensive test hosting and > a release cadence built on that. > > Or is the intention here to host software-factory as a RDO specific UI > backed by the CentOS pipeline ? > > > I think it might also be worthwhile making something a bit clearer as well. There is potentially two separate efforts in the works 1) To unify the hosting infrastructure and platform that all services that are part of RDO run on. The current proposal is for a new Openstack installation based on RDO itself to be deployed, in such a way that it's deployed and maintained transparent to the community (and indeed, conductive to community involvement). This has the potential to not only be useful for the RDO project, but potentially the CentOS project as well. The details of this are still being worked out, and I hope to speak more about this when I have something more concrete.

2) The potential to run an instance of the Software Factory software to leverage it as a workflow control tool for the development shipping process of RDO itself. This discussion I think was the original purpose of this thread, and is not tied to the final outcome of 1)

Hope this helps. Regards, Graeme -- Graeme Gillies Principal Systems Administrator Openstack Infrastructure Red Hat Australia _______________________________________________ Rdo-list mailing list Rdo-list(a)redhat.com https://www.redhat.com/mailman/listinfo/rdo-list To unsubscribe: rdo-list-unsubscribe(a)redhat.com