Ah, I wasn't personally aware of the distgit repos not mapping 1:1 with package name. I agree that there is something to fix here then. Thanks for clarifying

On Fri, Apr 6, 2018 at 11:00 AM, Stef Walter <stefw@redhat.com> wrote:
Three reasons:

 * dist-git repositories do not necessarily produce a package of the
   same name. They usually do.

 * dist-git repositories can produce multiple sub-packages, as you can
   see in the case of cockpit. Any and all of those sub-packages should
   be tested and/or gated by tests that live in that dist-git repo.

In this case there are 5 sub-packages from the cockpit dist-git repo in
the Atomic Host that should trigger the pipeline.

Although I understand current effort is on the All Package Pipeline, I
believe this might still be a problem going forward:

 * There is no defined way to discover which dist-git repos to look for
   tests related to a certain package in the standard test
   specification.

Does that make sense? Dominik, did you have any ideas on how we might
fix this?

Cheers,

Stef

On 06.04.2018 15:22, Johnny Bieren wrote:
> I'm not sure what you mean by fix it. If it isn't part of atomic host,
> shouldn't it not be tested by the atomic host pipeline?
>
> Cockpit *is* picked up by the All Package Pipeline. You can see one
> commit from cockpit here
> https://jenkins-continuous-infra.apps.ci.centos.org/blue/organizations/jenkins/upstream-fedora-f28-pipeline/detail/upstream-fedora-f28-pipeline/375/pipeline
>
> Unfortunately, that build is many builds ago so the artifacts have been
> discarded so you cannot see the package test logs anymore. But, wouldn't
> this be the solution, not having the atomic pipeline test it? 
>
> The integration with pagure for the All Packages Pipeline is still WIP I
> believe
>
> On Fri, Apr 6, 2018 at 7:35 AM, Stef Walter <stefw@redhat.com
> <mailto:stefw@redhat.com>> wrote:
>
>     On 03.04.2018 15:57, Johnny Bieren wrote:
>     > Hey all,
>     >
>     > To clear up some of the questions on this thread:
>     >
>     > - Why no cockpit testing? 
>     >   "cockpit" as a package name is not
>     > in https://pagure.io/fedora-atomic/blob/f27/f/fedora-atomic-host-base.json
>     <https://pagure.io/fedora-atomic/blob/f27/f/fedora-atomic-host-base.json>
>     > for example, so if the distgit repo name is "cockpit" the commit by
>     > design would not have been picked up by the Atomic Pipeline.
>
>     How do we fix that?
>
>     I guess this gap is because "Discovery" section of
>
>     https://fedoraproject.org/wiki/Changes/InvokingTests
>     <https://fedoraproject.org/wiki/Changes/InvokingTests>
>
>     Does not address how to find the dist-git repo for a given subpackage?
>
>     Stef
>
>
>     >
>     > - Atomic Pipeline doesn't test f28?
>     >   The Atomic pipeline is currently set up for f26 and f27. I have been
>     > working on getting it working for f28, but I am hitting a bug and
>     > haven't resolved it yet, so it is not ready yet.
>     >
>     > - What is the All Packages Pipeline
>     >
>     (https://jenkins-continuous-infra.apps.ci.centos.org/view/Fedora%20All%20Packages%20Pipeline/
>     <https://jenkins-continuous-infra.apps.ci.centos.org/view/Fedora%20All%20Packages%20Pipeline/>)
>     > ?
>     >   As alluded to, this is different than the Atomic pipelines. The
>     Atomic
>     > pipeline does the following: trigger on distgit commit to a repo of
>     > interest (see question 1 for package lists), build rpm, compose
>     ostree,
>     > compose qcow2 atomic image with ostree, boot sanity tests, package
>     > tests, atomic host integration tests, openshift e2e tests, report all
>     > results from all stages on fedmsg under org.centos.prod.ci.pipeline.*
>     > topics. The All Packages Pipelines do the following: trigger on
>     distgit
>     > commit to any package to branch [fXX | master] AND the repo contains
>     > standard-test-roles defined tests in that branch, build the rpm in
>     koji,
>     > construct cloud qcow2 image while injecting the rpm at build time, run
>     > package tests, report all results from all stages on fedmsg under
>     > org.centos.prod.allpackages.pipeline.* topics.
>     >
>     > - Is the All Packages Pipeline production ready?
>     >   I think it is. If you look
>     >
>     here https://jenkins-continuous-infra.apps.ci.centos.org/view/all/job/upstream-fedora-f28-pipeline/
>     <https://jenkins-continuous-infra.apps.ci.centos.org/view/all/job/upstream-fedora-f28-pipeline/>
>     > it is passing when it should. There even seem to be some legitimate
>     > package test failures found. If you go back in the build history, you
>     > will likely find some bugs that have since been fixed, but most of the
>     > latest builds are legit. We still need the linkage setup between these
>     > fedmsgs and pagure, though, to the best of my knowledge.
>     >
>     > If I can clear anything else up, feel free to ask.
>     > Best,
>     > Johnny Bieren
>     >
>     > On Mon, Mar 26, 2018 at 4:21 PM, Dominik Perpeet
>     <dperpeet@redhat.com <mailto:dperpeet@redhat.com>
>     > <mailto:dperpeet@redhat.com <mailto:dperpeet@redhat.com>>> wrote:
>     >
>     >     On 03/26/2018 05:34 PM, Ari LiVigni wrote:
>     >>
>     >>
>     >>     -== @ri ==-
>     >>
>     >>     On Mon, Mar 26, 2018 at 10:40 AM, Pierre-Yves Chibon
>     >>     <pingou@pingoured.fr <mailto:pingou@pingoured.fr>
>     <mailto:pingou@pingoured.fr <mailto:pingou@pingoured.fr>>> wrote:
>     >>
>     >>
>     >>
>     >>         For one, I'd need a green light that the allpackages pipeline
>     >>         is running as
>     >>         expected and ready for production use. Then I'll see to
>     adjust
>     >>         the tools for
>     >>         this new pipeline.
>     >>         This does mean we will announce it broadly and that we can
>     >>         expect its load to
>     >>         increase as more people opt-in.
>     >>
>     >
>     >     I agree. Let's make sure it's stable and production ready, then
>     >     announce it and help people when the inevitable issues and
>     questions
>     >     arise. :)
>     >
>     >>
>     >>         Do we want to discontinue the Atomic CI one? I thought the
>     >>         idea was to have both
>     >>         running (see what I said above about the two pipelines being
>     >>         complementary).
>     >>
>     >>
>     >>     Let's have that discussion when Johnny is back from PTO.  I am
>     >>     fine having both if it makes sense and there is no duplication,
>     >>     which I don't think there is since in the atomic one 
>     >>
>     >>
>     >>
>     >     I think both pipelines have merit, since they test different
>     >     deliverables. I would prefer to keep the Atomic one around,
>     but also
>     >     have the "regular" one once that is stable.
>     >
>     >     -Dominik
>     >
>     >
>     >     _______________________________________________
>     >     CI mailing list -- ci@lists.fedoraproject.org
>     <mailto:ci@lists.fedoraproject.org>
>     >     <mailto:ci@lists.fedoraproject.org
>     <mailto:ci@lists.fedoraproject.org>>
>     >     To unsubscribe send an email to ci-leave@lists.fedoraproject.org
>     <mailto:ci-leave@lists.fedoraproject.org>
>     >     <mailto:ci-leave@lists.fedoraproject.org
>     <mailto:ci-leave@lists.fedoraproject.org>>
>     >
>     >
>     >
>     >
>     > _______________________________________________
>     > CI mailing list -- ci@lists.fedoraproject.org
>     <mailto:ci@lists.fedoraproject.org>
>     > To unsubscribe send an email to ci-leave@lists.fedoraproject.org
>     <mailto:ci-leave@lists.fedoraproject.org>
>     >
>
>
>
>
> _______________________________________________
> CI mailing list -- ci@lists.fedoraproject.org
> To unsubscribe send an email to ci-leave@lists.fedoraproject.org
>