A few packages use eln branches to handle the differences between Fedora- and RHEL-style builds. When PRs are opened against such eln branches, the CI scratch build (in the non-Zuul case) is run against rawhide rather than the eln target, which returns inaccurate results.
eln
Recent example: https://src.fedoraproject.org/rpms/golang/pull-request/234
CI for PRs opened against an eln branch should use the eln target instead.
Hi, Could you confirm that jobs triggered by Zuul are not affected ?
There aren't that many eln branches, so I'd have to check, but rawhide PRs of Zuul-enabled packages do scratch builds for both rawhide and eln.
Would https://github.com/fedora-ci/dist-git-pipeline/pull/64 fix this?
https://github.com/fedora-ci/dist-git-build-pipeline/pull/45 https://github.com/fedora-ci/dist-git-trigger/pull/6
For testing purposes, we currently have two ELN PRs open:
https://src.fedoraproject.org/rpms/lorax-templates-rhel/pull-request/24 https://src.fedoraproject.org/rpms/pcs/pull-request/60
Scratch builds are now against the eln-candidate target as requested, but e.g. running the dist-git test against lorax-templates-rhel, a weird mixture of rawhide and ELN packages were installed. What defines the testing-farm-tag-repository and test-artifacts yum.repos.d in TMT?
testing-farm-tag-repository
test-artifacts
Also, https://github.com/fedora-ci/installability-pipeline/pull/59 seems necessary for the installability pipeline to trigger.
/cc @mvadkert
Added https://github.com/fedora-ci/mini-tps/pull/90 to match.
Looks like beakerlib is needed; filed https://github.com/minimization/content-resolver-input/pull/1490 to add it to ELN.
Filed more PRs, here's a list of what's now pending: https://github.com/fedora-ci/mini-tps/pull/90 https://github.com/fedora-ci/installability-pipeline/pull/59 https://github.com/fedora-ci/installability-trigger/pull/7 https://github.com/teemtee/tmt/pull/4042
The dist-git-build, dist-git, and installability pipelines appear to be working properly on ELN PRs, albeit with a limited sample set so far.
Metadata Update from @yselkowitz: - Issue status updated to: Closed (was: Open)