Comments
adamwill
Several missing rebuilds:
Dependencies of other packages that would be BROKEN by the tested packages:
package: gala-8.4.0^20260119.gitb856360-1.fc44.x86_64 from https://kojipkgs.fedoraproject.org/repos/f44-build/latest/x86_64
libmutter-17.so.0()(64bit)
libmutter-clutter-17.so.0()(64bit)
libmutter-cogl-17.so.0()(64bit)
libmutter-mtk-17.so.0()(64bit)
package: gala-devel-8.4.0^20260119.gitb856360-1.fc44.x86_64 from https://kojipkgs.fedoraproject.org/repos/f44-build/latest/x86_64
pkgconfig(libmutter-17) < 50
pkgconfig(libmutter-17) >= 49
package: gnome-kiosk-49.0-4.fc44.x86_64 from https://kojipkgs.fedoraproject.org/repos/f44-build/latest/x86_64
libmutter-17.so.0()(64bit)
libmutter-clutter-17.so.0()(64bit)
libmutter-cogl-17.so.0()(64bit)
libmutter-mtk-17.so.0()(64bit)
package: gnome-shell-extension-background-logo-49.0-3.fc44.noarch from https://kojipkgs.fedoraproject.org/repos/f44-build/latest/x86_64
gnome-shell(api) = 49
gnome-kiosk causes the openQA test failures as it's a dependency of anaconda, but they should all be fixed. This keeps happening; does mclazy need an update to know about these?
This one seems a bit better than previous attempts - at least tests that update an existing install all passed, and the live image build/install/boot tests (which boot to a graphical login manager) passed. But the tests where we build a Cloud image then boot from it, and the tests where we build a network install image, install from it with a side repo containing the package under test, and boot the installed system both failed. It seems we get no tty on vt1 in each case. We do get ttys on the other VTs, it's only vt1 that's bad.
not sure if openQA failures were actually real, they may have been blips, but in any case, the update was replaced with -19 which passed.
soname issue, was replaced by later updates that ultimately passed gating.
procps-ng contained an unannounced soname bump, see https://bodhi.fedoraproject.org/updates/FEDORA-2026-3f8f5e81e1#comment-4534677 .
This bumped an soname - libproc2.so.0 to libproc2.so.1 - and broke several deps:
Dependencies of other packages that would be BROKEN by the tested packages:
package: cpu-x-5.4.0-3.fc44.x86_64 from https://kojipkgs.fedoraproject.org/repos/f44-build/latest/x86_64
libproc2.so.0()(64bit)
libproc2.so.0(LIBPROC_2)(64bit)
package: igt-gpu-tools-2.2-2.fc44.x86_64 from https://kojipkgs.fedoraproject.org/repos/f44-build/latest/x86_64
libproc2.so.0()(64bit)
libproc2.so.0(LIBPROC_2)(64bit)
package: lxqt-session-2.3.0-2.fc44.x86_64 from https://kojipkgs.fedoraproject.org/repos/f44-build/latest/x86_64
libproc2.so.0()(64bit)
libproc2.so.0(LIBPROC_2)(64bit)
package: xpra-1:6.4.3-1.fc44.x86_64 from https://kojipkgs.fedoraproject.org/repos/f44-build/latest/x86_64
libproc2.so.0()(64bit)
libproc2.so.0(LIBPROC_2)(64bit)
This should have been announced to devel@ and communicated to maintainers of depending packages to allow them to be rebuilt. See https://docs.fedoraproject.org/en-US/fesco/Updates_Policy/#_rawhide :
"When a proposed update contains an ABI or API change: notify a week in advance both the devel list and maintainers directly (using the packagename-maintainers@fedoraproject.org alias) whose packages depend on yours to rebuild or offer to do these rebuilds for them.
Use a side-tag when dealing with mass builds of many packages, so they can land at the same time. See Rawhide Gating/Multi Builds."
This update has been unpushed.
For the record, https://github.com/cockpit-project/cockpit/pull/22828 is intended to fix the update problem. I'm gonna unpush this update since we know it has that bug.
Ugh, still something off with login-manager. Will look at it tomorrow.
No, it's not listed as fixing it, but Martin's making other changes to the package that are getting automatically built, while we're trying to figure that one out.
I've had this untagged because it was built against the old poppler. I've got a rebuild running now. https://forge.fedoraproject.org/releng/tickets/issues/13194
Still failing in nc TLS stuff, but it's a bit different now:
:: [ 20:30:08 ] :: [ BEGIN ] :: Starting netcat :: actually running 'nc -l 1125 -c mockserver.sh &'
nc: tls configuration failed (private/public key mismatch)
Still broken when tested with FEDORA-2026-0b2b8265d2 .
Note that's without the fedora-release change, as that's still gated too. Again, if both fedora-release and systemd need to go together, they should be in a combined update. Let me know if you want me to test them together again.
Still no tty on vt2, vt3 or vt6 in openQA testing.
Looks like this was missing a rebuild of kstars, per rmdepcheck:
Dependencies of other packages that would be BROKEN by the tested packages:
package: kstars-1:3.8.0-3.fc44.x86_64 from https://kojipkgs.fedoraproject.org/repos/f44-build/latest/x86_64
libopencv_calib3d.so.412()(64bit)
libopencv_core.so.412()(64bit)
libopencv_imgproc.so.412()(64bit)
I figured out the problem here and sent a fix. That was a pretty fun one...
This seems to be easily reproducible manually - just run Cockpit and install some updates (it doesn't have to be the special one openQA sets up in its test, I hit it just updating a day-old Rawhide VM). Console log messages:
Object { problem: null, name: "org.freedesktop.DBus.Error.UnknownInterface", message: "Unknown interface 'org.freedesktop.systemd1.Service'.", toString: toString() }
overview.js:69:70827
Transaction has an invalid timespec=2026-01-29T12:57:49.112827-08 updates.js:45:39523
TypeError: can't access property "packages", i[0] is undefined
TR http://localhost:9090/cockpit/@localhost/updates/updates.js:160
lb http://localhost:9090/cockpit/@localhost/updates/updates.js:6
yk http://localhost:9090/cockpit/@localhost/updates/updates.js:8
xk http://localhost:9090/cockpit/@localhost/updates/updates.js:8
DT http://localhost:9090/cockpit/@localhost/updates/updates.js:8
gd http://localhost:9090/cockpit/@localhost/updates/updates.js:8
Eu http://localhost:9090/cockpit/@localhost/updates/updates.js:8
bk http://localhost:9090/cockpit/@localhost/updates/updates.js:8
gg http://localhost:9090/cockpit/@localhost/updates/updates.js:1
fg http://localhost:9090/cockpit/@localhost/updates/updates.js:1
updates.js:8:161
Uncaught TypeError: can't access property "packages", i[0] is undefined
TR http://localhost:9090/cockpit/@localhost/updates/updates.js:160
lb http://localhost:9090/cockpit/@localhost/updates/updates.js:6
yk http://localhost:9090/cockpit/@localhost/updates/updates.js:8
xk http://localhost:9090/cockpit/@localhost/updates/updates.js:8
DT http://localhost:9090/cockpit/@localhost/updates/updates.js:8
gd http://localhost:9090/cockpit/@localhost/updates/updates.js:8
Eu http://localhost:9090/cockpit/@localhost/updates/updates.js:8
bk http://localhost:9090/cockpit/@localhost/updates/updates.js:8
gg http://localhost:9090/cockpit/@localhost/updates/updates.js:1
fg http://localhost:9090/cockpit/@localhost/updates/updates.js:1
updates.js:160:7569
they look like they're a bit minified, but maybe it's still enough? If not I can try to install unminified cockpit somehow...
openQA is showing the same "Ooops!" problem when installing updates as the Rawhide update. Notably, F43 does not have the PackageKit dnf5 backend change, so it looks like that can't be the cause.
Looks like the revert was incomplete. I'm trying a scratch build with more reversion in split-files.py.