On 12/15/2016 07:13 AM, Niels de Vos wrote:
On Thu, Dec 15, 2016 at 05:20:26PM +0530, Nigel Babu wrote:
Hello folks,
This is a request for feedback for this plan to build a pipeline that will go
from a commit to builds that are tested and signed. While the
nightly.gluster.org site will be up by the time 3.10 is released, we will not
yet be automating releases with this pipeline by then.
Could you explain what distributions, versions and architectures you can
cover with this? I suspect that the different distributions that provide
packages for the glusterfs packages have a broader range than what we
can test in our environment?
Also, builds for distributions normally need to be done in the build
system of those distributions. They will not accept builds from external
environments... I am not sure if/how this can become a replacement for
packages that are part of the distributions.
To me this looks like a lot of duplication of the plans with the builds
done through the CentOS CI and have the different CentOS SIGs consume
those builds for their testing. Has there been any thought about running
tests against other projects that use Gluster?
We've already discussed that the CentOS Build System can be used for
doing nightly builds by bots, and have these progress from testing to
release repositories. At least for the RPMs that are used by CentOS (and
possibly RHEL) users, I would much prefer this approach.
AIUI, Nigel is planning to tackle a few things:
1) automate submitting builds to Launchpad, for our Launchpad PPA.
This doesn't tie in to any Ubuntu "official" packages of GlusterFS.
IIRC he knows how we might to the equivalent (or what I think of
as the equivalent) of scratch builds in Launchpad that won't
affect the genuine PPA builds but we can refine and debug the
builds before a release if there are problems.
2) automate submitting builds to SuSE Build System. Similar to
Launchpad, but generally easier because they are RPMs.
3) automate Debian builds for d.g.o.
Nigel may have grander schemes above and beyond that, and that's fine by
me, but none of the above, or his grander schemes, have any real overlap
(AFAIK) with what we're doing with the CentOS CI.
Nigel will correct me if I'm wrong.
--
Kaleb
_______________________________________________
Gluster-devel mailing list
Gluster-devel@xxxxxxxxxxx
http://www.gluster.org/mailman/listinfo/gluster-devel
