It seems that we missed some files during revert of all the bodhi changes. Let's
delete them for now and revert the commit later if needed.
Signed-off-by: Michal Konečný <mkonecny@redhat.com>
I forgot to update imports when I renamed the file during migration to
templates. This caused the non-template file to be imported instead, which was
the reason of the error.
Signed-off-by: Michal Konečný <mkonecny@redhat.com>
Still not working. ;(
So, next time we think we have this, we should push it out and make sure
to push one updates-testing or something to actually make sure it works!
Signed-off-by: Kevin Fenzi <kevin@scrye.com>
There are some values that need to be filled in the pungi_general.conf. Let's move it
to templates, so it could be processed as every other template.
Signed-off-by: Michal Konečný <mkonecny@redhat.com>
This reverts commit bf345042d2.
This is breaking all updates composes.
pungi.compose.status.change -- pungi-koji compose of [[ release.id_prefix.title() ]]-7-updates-20230429.0 started https://kojipkgs.fedoraproject.org/compose/updates/[[ release.id_prefix.title() ]]-7-updates-20230429.0
somehow it's not expanding right so reverting it.
Right now we are pushing a pointless armhfp variant in f37/f38 updates.
(It only has noarch packages in it).
So, lets clean up the logic here for both rpm updates pushes and modular
pushes. We can drop atomic as we don't do that anymore.
After we push this out and confirm it's working for a day or two, we
should go in and remove the bogus armhfp dirs from the master mirrors.
Signed-off-by: Kevin Fenzi <kevin@scrye.com>
This converts the automated pushes cron template to just a simple file
thats always enabled. Since new bodhi has a way to 'freeze' frozen
releases, it won't push stable updates for any of those releases (but
will push the rest). So, we can just leave it always enabled as long as
we properly freeze/unfreeze releases.
Since it's just a simple file now, move it to one and change it from a
template.
Signed-off-by: Kevin Fenzi <kevin@scrye.com>
Move it to here instead of having systemd-devel fake require
systemd-pam.
This should help with flatpak runtime packaging so that we can avoid
having to ship systemd-pam in the flatpak container.
`wine*` affected a package which didn't need to be multilib (`wineglass`). At
the same time, we need to match not only all subpackages of `wine`, but also
separate packages like `wine-dxvk`. The new approach should be a little better.
Signed-off-by: Kamil Páral <kparal@redhat.com>
This has been broken since basically forever.
The bodhi keytab is owner by apache and expects to be run under that
user. Two reasons we never noticed/fixed this:
Until recently it was failing with a error trying to talk to pagure.io's
api. So I just thought it was too slow and needed to be optomized.
When I tweaked pagure's httpd config for the DDoS we had the other day I
fixed this on the server end, then it got to the auth error.
Signed-off-by: Kevin Fenzi <kevin@scrye.com>
