side tag created:

$koji add-tag f39-python --parent=f39-build --arches=i686,x86_64,aarch64,ppc64le,s390x
$koji add-target f39-python f39-python
$koji edit-tag f39-python -x 'mock.use_bootstrap=0'

Signing PR: https://pagure.io/fedora-infra/ansible/pull-request/1477

Thank you.

Signing pr merged and deployed.

Note: Before merging this side tag, we will need to re-enable mock bootstrap in it and verify we can still build (e.g. a scratchbuild), as we will rebuild dnf and friends from dist-git which will likely reintroduce https://pagure.io/releng/issue/11441 unless https://pagure.io/releng/issue/11446 is resolved by then.

This night, the repo wasn't regenerated for hours until I did it manually in the morning.

Could @thrnciar please gain the repo permission as well?

$ koji regen-repo f39-python
2023-06-15 10:42:35,136 [ERROR] koji: ActionNotAllowed: repo permission required (logged in as thrnciar)

Odd, kojira should be doing this for you... no need for manual regens...

It does seem to be?

Jun 15 15:12:33 koji02.iad2.fedoraproject.org kojira[943]: 2023-06-15 15:12:33,355 [INFO] {943} koji.repo.regen:951 Tag needs regen: f39-python
Jun 15 15:12:45 koji02.iad2.fedoraproject.org kojira[943]: 2023-06-15 15:12:45,723 [INFO] {943} koji.repo.regen:1071 Created newRepo task 102179509 for tag 68571 (f39-python), expired for 23.3 sec

It works, but not reliably and sometimes it's very slow. Tonight, I built scipy at 22:50:01 UTC and it only appeared in the repo after I run regen-repo in the morning.

Odd. It does have a gigantic pile of sidetags to update, but it should mostly be doing pretty well.

Anyhow, I have added 'repo' perms for thrnciar in the mean time.

I guess we could keep this to try and figure out what could be going on with kojira, or just close it and deal with that some other time...

Could you please merge the side tag now?

I guess we could keep this to try and figure out what could be going on with kojira, or just close it and deal with that some other time...

It turned out to behave quite well since.

Could you please merge the side tag now?

Actually, please coordinate with @adamwill.

If it's ready today I can do it, if it's after I have gone to sleep I'll leave it to @humaton :)

I'm working on all the fallout openQA has found, I will comment here if I get it all to go green today. Currently fixing dogtag-pki.

Still working...latest thing is that anaconda exit (including quit from a live install, or reboot at the end of a regular install) fails with:

RuntimeError: preexec_fn not supported at interpreter shutdown

edit: this appears to be https://github.com/python/cpython/pull/104826 , which was done to address https://github.com/python/cpython/issues/104690 .

edit 2: testing a fix for this. In "production" (the f39-python tag), because how else do we test things?!

KDE live build fails because kpat requires libfreecell-solver, libfreecell-solver requires freecell-solver-data, and freecell-solver-data hasn't been rebuilt.

edit: oh fun, it's a giant dependency hole. looks like freecell-solver-data requires python3-freecell_solver, which hasn't been rebuilt yet because at least one of its deps isn't rebuilt yet. Its deps are:

BuildRequires: python3dist(coverage) >= 4.0
BuildRequires: python3dist(hacking) >= 0.12.0
BuildRequires: python3dist(openstackdocstheme)
BuildRequires: python3dist(oslotest) >= 1.10.0
BuildRequires: python3dist(pbr)
BuildRequires: python3dist(pbr) >= 2.0
BuildRequires: python3dist(python-subunit) >= 0.0.18
BuildRequires: python3dist(setuptools)
BuildRequires: python3dist(testtools) >= 1.4.0
BuildRequires: python3dist(sphinx)

looks like hacking hasn't had a build attempted yet (presumably another layer of deps there), pbr is in 'bootstrap' state (did someone forget to do the post-bootstrap build?), and python-testtools failed to build.

I don't think I have time to unpick all of that tonight. We can do it tomorrow, or we can temporarily drop kpat from the KDE live, I guess.

did someone forget to do the post-bootstrap build (of pbr)?

It is not ready yet. There are 24 packages like this. Don't worry about it, we'll commit the debootsrap commits before the mass rebuild.

Building python-freecell_solver in https://koji.fedoraproject.org/koji/taskinfo?taskID=102943948 via https://src.fedoraproject.org/rpms/python-freecell_solver/c/0db02a9746c2157869ec538272b8195ff6883553?branch=rawhide

EDIT and freecell-solver in https://koji.fedoraproject.org/koji/taskinfo?taskID=102944214

Another issue in FreeIPA: it uses the removed ssl.match_hostname(). This breaks the FreeIPA upgrade test. I think it can probably be switched to use the implementation from urllib3 (urllib3.util.match_hostname()) - freeipa doesn't currently have a direct dependency on urllib3, but it uses requests, so it already depends on it indirectly. I'd do this, but it's past 11pm here, if nobody else does it before I wake up, I'll look at it.

kpat installs now and so does the entire @kde-apps

OK, so just this new FreeIPA issue remaining.

FreeIPA issue fixed.

Please wait until https://koji.fedoraproject.org/koji/taskinfo?taskID=102951115 is finished and proceeded with the merge.

Please wait until https://koji.fedoraproject.org/koji/taskinfo?taskID=102951115 is finished and proceeded with the merge.

That build has finished.

After the merge, let's tag in samba-4.18.3-5.fc39 manually. Somebody built it again in rawhide in the meantime and I don't want to hold off the merge for another hour or so.

Thank you!