Python 3.15.0b4 and extension modules rebuilt for it
Automatic update for criu-4.2.1-1.fc45.
Automatic update for osbuild-187-1.fc45.
Automatic update for pygobject3-3.57.0-1.fc45.
Automatic update for unbound-1.25.1-4.fc45.
Automatic update for python-pandas-3.0.5-2.fc45.
Updates may require up to 24 hours to propagate to mirrors. If the following command doesn't work, please retry later:
sudo dnf upgrade --refresh --advisory=FEDORA-2026-2551d0d66bPlease log in to add feedback.
This update's test gating status has been changed to 'waiting'.
This update has obsoleted criu-4.2.1-1.fc45, and has inherited its bugs and notes.
This update has obsoleted osbuild-187-1.fc45, and has inherited its bugs and notes.
This update has obsoleted pygobject3-3.57.0-1.fc45, and has inherited its bugs and notes.
This update has obsoleted unbound-1.25.1-4.fc45, and has inherited its bugs and notes.
Packages still building: llvm, vtk.
This update's test gating status has been changed to 'waiting'.
This update's test gating status has been changed to 'failed'.
All rpmdepchecks fail with:
OSError: [Errno 7] Argument list too long: 'dnf'
Because this update is too big.
churchyard edited this update.
New build(s):
Karma has been reset.
churchyard edited this update.
New build(s):
Removed build(s):
Karma has been reset.
This update has obsoleted python-pandas-3.0.5-2.fc45, and has inherited its bugs and notes.
This update's test gating status has been changed to 'waiting'.
This update's test gating status has been changed to 'failed'.
This update's test gating status has been changed to 'waiting'.
This update's test gating status has been changed to 'failed'.
Other openQA tests are failing because this update is so huge it takes longer than expected to download. Will bump some timeouts...
churchyard edited this update.
New build(s):
Karma has been reset.
churchyard edited this update.
New build(s):
Removed build(s):
Karma has been reset.
Ugh, even with a bumped timeout it doesn't work. There seems to be some kind of phenomenon where when an update is X big, the downloads just get...stuck, like trying to download that many packages across that many tests at once chokes something up and it all just grinds to a halt.
The last time this happened I worked around it by special-casing openQA to test the update's side tag directly, rather than the NVRs. This is less good because it's vulnerable to the contents of the side tag changing. But we might have to do the same here to get some tests through. I'm going to look at generalizing that workaround a bit and maybe see if I can have it test a specific state of the side tag buildroot, rather than just the 'latest', see if that helps with the sync issues.
So now I got openQA working properly, it found a real problem. This update includes pygobject3-3.57.0-2.fc45 , a rebuild of pygobject3-3.57.0-1.fc45 , but that build is actually broken and its update is gated. It causes anaconda shutdown to fail ~50% of the time.
Unless upstream comes up with a fix for that soon, you'll need to revert the 3.57.0 bump in pygobject3 git and do a pygobject3-3.56.3-6.fc45 build instead.
This update's test gating status has been changed to 'waiting'.
This update's test gating status has been changed to 'failed'.
is pygobject3 the only real problem?
churchyard edited this update.
New build(s):
Karma has been reset.
churchyard edited this update.
New build(s):
Removed build(s):
Karma has been reset.
churchyard edited this update.
New build(s):
Removed build(s):
Karma has been reset.
Detectable by openQA, yeah. There may be some races with other builds of packages in the update that we'd only discover on attempting to push stable.
This update's test gating status has been changed to 'waiting'.
This update's test gating status has been changed to 'failed'.
churchyard edited this update.
New build(s):
Removed build(s):
Karma has been reset.
churchyard edited this update.
New build(s):
Removed build(s):
Karma has been reset.
pygobject3 reverted, all Perl 5.44 rebuild races rebuilt
Some of the gating fedora-ci.koji-build.tier0.functional tests failed during guest-setup.
If this was a releng-managed side tag, we would not have run those at all, so I am inclined to waive them.
Waiting for the rest of the tests to finish.
The failing Fedora CI tests are I believe because of the system's current behavior of trying to install all packages from the update - they hit a bunch of internal conflicts. See this log for e.g. CC @lecris for confirmation.
openQA failures so far are the same as before, plus FAS appears to be down which causes desktop_browser to fail - I'll deal with that somehow as it's obviously also affecting other updates.
ah, so, hmm, yeah. It's still hitting the anaconda grey screen issue because of one of those annoying sync problems that happen when testing the side tag rather than the exact NVR list.
for the side tag thing I made the scheduler find the latest repo for the side tag at the time of scheduling and have the tests use that. But it looks like you didn't force a repo re-gen when updating the update, or anything, so the side tag repo at the time you updated the update did not have the updated builds in it - note https://kojipkgs.fedoraproject.org/repos/f45-build-side-144323/6678939/x86_64/pkglist still has the 3.57.0-2 build in it. This is why we can't just have openQA test all side tag updates via the side tag, unfortunately.
I'll force a regen of the side tag repo and reschedule the tests manually.
Agh, KDE live installs now fail because this update includes anaconda-45.13-2 , which has the same bug as anaconda-45.13-1 which was gated. You either need to wait for a new anaconda+anaconda-webui update and rebuild that (not sure when they're planning to do it), or revert anaconda to 45.12 and rebuild that.
churchyard edited this update.
New build(s):
Removed build(s):
Karma has been reset.
This update's test gating status has been changed to 'waiting'.
This update's test gating status has been changed to 'failed'.
churchyard edited this update.
New build(s):
Removed build(s):
Karma has been reset.
This update's test gating status has been changed to 'passed'.
This update has been submitted for stable by bodhi
So I've now manually waived the remaining failures - the magic URL is https://waiverdb.fedoraproject.org/api/v1.0/waivers/new for future reference, no I don't know why an interactive page is under /api - but we're still gated. I think this is because you can't waive missing results, only failed results (I'm 95% sure I figured this out in the past, the code is in greenwave and it's kinda twisty so I can't point to it right now, will confirm in the morning).
The tests are missing because Fedora CI refuses to run some tests on updates over a certain size. So there isn't a lot we can do at this point without changing the CI pipelines or changing greenwave. We may just need releng to merge the tag manually.
wait, huh? OK. I dunno how that worked, but great.