[HEADS UP] ovift-engine: repository closure check on check-patch
Hi, with the following commits: - https://gerrit.ovirt.org/71553 - https://gerrit.ovirt.org/71623 a minimal repository closure check is added to check-patches.sh. Instead of performing a full blown repository closure, I limited the check to the built rpms in oreder to reduce the risk of hitting repository closure errors due to different projects. This should help dropping the long time failing repository closure jobs which are struggling to keep the pace with the changes we introduce in master. It should also help avoiding to merge patches which will break the repository and the plain rpm install flow. The second patch switch from ovirt-master-release to experimental repo. Here I need infra input, not sure experimental is the desired repo, deisgned to replace ovirt-master-snapshot. Thanks, -- Sandro Bonazzola Better technology. Faster innovation. Powered by community collaboration. See how it works at redhat.com
On Fri, Feb 24, 2017 at 4:59 PM, Sandro Bonazzola <sbonazzo@redhat.com> wrote:
Hi, with the following commits: - https://gerrit.ovirt.org/71553 - https://gerrit.ovirt.org/71623
a minimal repository closure check is added to check-patches.sh. Instead of performing a full blown repository closure, I limited the check to the built rpms in oreder to reduce the risk of hitting repository closure errors due to different projects.
This should help dropping the long time failing repository closure jobs which are struggling to keep the pace with the changes we introduce in master. It should also help avoiding to merge patches which will break the repository and the plain rpm install flow.
Forgot to mention I recomend to steal the check for your own project reducing repository closure issues as well :-)
The second patch switch from ovirt-master-release to experimental repo. Here I need infra input, not sure experimental is the desired repo, deisgned to replace ovirt-master-snapshot.
Thanks, -- Sandro Bonazzola Better technology. Faster innovation. Powered by community collaboration. See how it works at redhat.com
-- Sandro Bonazzola Better technology. Faster innovation. Powered by community collaboration. See how it works at redhat.com
Sorry, but I really don't think this is the right direction to go. This exposes check-patch to outside repos and may result in repo-stability-related failures after we've made great efforts to eliminate those. This kind of test should really be done at the OST stage IMO. בתאריך 24 בפבר׳ 2017 18:01, "Sandro Bonazzola" <sbonazzo@redhat.com> כתב:
On Fri, Feb 24, 2017 at 4:59 PM, Sandro Bonazzola <sbonazzo@redhat.com> wrote:
Hi, with the following commits: - https://gerrit.ovirt.org/71553 - https://gerrit.ovirt.org/71623
a minimal repository closure check is added to check-patches.sh. Instead of performing a full blown repository closure, I limited the check to the built rpms in oreder to reduce the risk of hitting repository closure errors due to different projects.
This should help dropping the long time failing repository closure jobs which are struggling to keep the pace with the changes we introduce in master. It should also help avoiding to merge patches which will break the repository and the plain rpm install flow.
Forgot to mention I recomend to steal the check for your own project reducing repository closure issues as well :-)
The second patch switch from ovirt-master-release to experimental repo. Here I need infra input, not sure experimental is the desired repo, deisgned to replace ovirt-master-snapshot.
Thanks, -- Sandro Bonazzola Better technology. Faster innovation. Powered by community collaboration. See how it works at redhat.com
-- Sandro Bonazzola Better technology. Faster innovation. Powered by community collaboration. See how it works at redhat.com
_______________________________________________ Infra mailing list Infra@ovirt.org http://lists.ovirt.org/mailman/listinfo/infra
Il 24/Feb/2017 19:37, "Barak Korren" <bkorren@redhat.com> ha scritto: Sorry, but I really don't think this is the right direction to go. This exposes check-patch to outside repos and may result in repo-stability-related failures after we've made great efforts to eliminate those. This kind of test should really be done at the OST stage IMO. OST come into play too late, with patch already merged and dep tree broken. It will avoid the rpm to be published but it will fail the subsequent ost hiding other issues. Any suggestion on how to avoid it? Run OST in check patch? I haven't merged the patches becauseI wanted feedback like this, thanks Barack. בתאריך 24 בפבר׳ 2017 18:01, "Sandro Bonazzola" <sbonazzo@redhat.com> כתב:
On Fri, Feb 24, 2017 at 4:59 PM, Sandro Bonazzola <sbonazzo@redhat.com> wrote:
Hi, with the following commits: - https://gerrit.ovirt.org/71553 - https://gerrit.ovirt.org/71623
a minimal repository closure check is added to check-patches.sh. Instead of performing a full blown repository closure, I limited the check to the built rpms in oreder to reduce the risk of hitting repository closure errors due to different projects.
This should help dropping the long time failing repository closure jobs which are struggling to keep the pace with the changes we introduce in master. It should also help avoiding to merge patches which will break the repository and the plain rpm install flow.
Forgot to mention I recomend to steal the check for your own project reducing repository closure issues as well :-)
The second patch switch from ovirt-master-release to experimental repo. Here I need infra input, not sure experimental is the desired repo, deisgned to replace ovirt-master-snapshot.
Thanks, -- Sandro Bonazzola Better technology. Faster innovation. Powered by community collaboration. See how it works at redhat.com
-- Sandro Bonazzola Better technology. Faster innovation. Powered by community collaboration. See how it works at redhat.com
_______________________________________________ Infra mailing list Infra@ovirt.org http://lists.ovirt.org/mailman/listinfo/infra
On 25 February 2017 at 17:57, Sandro Bonazzola <sbonazzo@redhat.com> wrote:
OST come into play too late, with patch already merged and dep tree broken.
check_patch is too early - before a repo has been composed. There is quite a bit of repo composition logic in the CI system, trying to imitate it with outside sources will probably lead to the wrong results. repo closure is a system-level test, it does not belong in check-patch IMO.
It will avoid the rpm to be published but it will fail the subsequent ost hiding other issues. Any suggestion on how to avoid it? Run OST in check patch?
Currently this is unavoidable for all system-level regressions, but we are making big efforts to close loops there as fast as possible. Change gating is the long-term technical solution for this. It will both prevent subsequent OST runs from failing, and will allow accurate regression source detection. We've made progress with in in recent months. You are welcome to go and review the patch that implements the change tracking logic, it is in your review queue.... Once we have change gating, up and running, we will be able to move to the 2nd stage of applying it to select projects pre-merge. Engine will most probably be one of the projects this is applied to.
I haven't merged the patches because I wanted feedback like this, thanks Barack.
Thanks for consulting with us. One nitpick - "Barack" is how the former POTUS spells it, I spell it "Barak". -- Barak Korren bkorren@redhat.com RHCE, RHCi, RHV-DevOps Team https://ifireball.wordpress.com/
participants (2)
-
Barak Korren -
Sandro Bonazzola