Commit graph

4,406 commits

Author SHA1 Message Date
a3ac35d94e setup_repos.py: re-use Koji session
All checks were successful
CI via Tox and perl / tox (pull_request) Successful in 1m25s
CI via Tox and perl / perl (pull_request) Successful in 3m30s
CI via Tox and perl / checkwiki (pull_request) Successful in 33s
I thought this wasn't possible as I didn't think the Koji client
library was thread-safe, but it seems like it actually is, as it
uses requests. I tested this out locally and it looks like it's
OK.

We noticed this "check whether this build actually ran for our
arch" code generated a lot of load on Koji when a 400+ package
qt6 update hit testing and both openQA instances span up a bunch
of jobs, many of which hit setup_repos simultaneously. I hope
this will mitigate it a bit in future.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-16 01:12:58 -07:00
4dc1786278 rmdepcheck: handle rdc-repos / rdc-compose split
All checks were successful
CI via Tox and perl / perl (pull_request) Successful in 3m12s
CI via Tox and perl / tox (pull_request) Successful in 1m18s
CI via Tox and perl / checkwiki (pull_request) Successful in 33s
AI Code Review / ai-review (pull_request_target) Successful in 23s
This adapts to the planned upstream split of rmdepcheck into
rdc-repos and rdc-compose:
quality/rmdepcheck#40
rdc-repos is the 'old' mode, still used for most cases.
rdc-compose is the new mode intended for EL cases, especially ELN;
we use it for ELN now.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-15 23:30:19 -07:00
d42b66b9de desktop_notifications: try to prevent "System Updates Installed" notifications
All checks were successful
CI via Tox and perl / tox (pull_request) Successful in 1m26s
CI via Tox and perl / perl (pull_request) Successful in 3m6s
CI via Tox and perl / checkwiki (pull_request) Successful in 33s
Because they started to appear and we don't want to see them.

Also remove "online-updates-timestamp" because that key doesn't exist anymore
(an error is printed saying the same, when you try to assign a value to it).

Related: quality/tickets#899
2026-05-14 12:06:02 -07:00
2a8dd8e5fc Update Japanese language selection needle for webUI
Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-14 12:05:12 -07:00
caeb10ae2e Update Firefox addon_install needles for a new release
Dunno why the rendering of this button always changes. Oh, well.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-14 12:04:31 -07:00
ad93b627ca Add current maps_info needles
The test currently always soft-fails because these are missing.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-14 12:03:34 -07:00
1e88d8d8a0 Maps: always use ctrl-f to search
Hitting ctrl-f selects all text in the search field, so we can
just use it every time. Simpler and more robust.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-14 11:53:56 -07:00
b9fa791bb2 setup_tap_static: don't assert the nmcli down
If the connection is down already for some reason, that's not
really a problem. Let's not fail.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-14 11:27:09 -07:00
255247b5a6 Install the Flathub version of Media Writer
All checks were successful
CI via Tox and perl / tox (pull_request) Successful in 1m19s
CI via Tox and perl / perl (pull_request) Successful in 3m11s
CI via Tox and perl / checkwiki (pull_request) Successful in 32s
AI Code Review / ai-review (pull_request_target) Successful in 1m6s
By default, Silverblue comes with a pre-installed version of Media
Writer which unfortunately does not cover the Flathub test case for
Media Writer.

This PR removes the pre-installed version, adds the Flathub repository
and installs the application from there as a preparatory step for
the actual Media Writer testing.

Thus, we can correctly report the results for the Flathub tests.
2026-05-13 09:51:17 -07:00
e0db5d67b7 toolbox: pass buildroot repo through to container
As well as all the others.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-12 18:46:15 -07:00
cdbb54e230 Narrow match area on gis_button_start_fedora needle for ELN
The button text is "Start Using Fedora ELN" on ELN. Let's just
narrow the match area to just "Start Using Fedora" so it matches
everywhere, instead of adding a variant.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-12 18:45:13 -07:00
848e7ef3af Add settings to check-needles list of apps_run needles
Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-12 18:02:08 -07:00
97057598f6 podman: tweak extra filter handling, skip a flaky test on Rawhide
Some checks failed
CI via Tox and perl / tox (pull_request) Failing after 1m29s
CI via Tox and perl / perl (pull_request) Successful in 3m10s
CI via Tox and perl / checkwiki (pull_request) Successful in 32s
AI Code Review / ai-review (pull_request_target) Successful in 55s
This makes the extra filter handling more sophisticated so we can
easily add and remove extra filters, and skips a test that's
currently flaky on Rawhide, see
https://bugs.passt.top/show_bug.cgi?id=202 .

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-12 17:52:02 -07:00
1d42b09dda _type_user_password: switch back to ASCII after typing
Otherwise, in a web UI, we leave the layout as native when typing
the root password and it doesn't come out as what we expect.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-12 17:36:43 -07:00
fd428212b0 g-i-s: suppress "last login" messages on ELN
On ELN PAM is configured to always print a "Last login" message
on login and that breaks our expected output for run_as_user.
We can't use runuser instead because with that we get warnings
about not being able to create /run/user/0/dconf.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-12 17:15:06 -07:00
ab71faa4c9 g-i-s: skip third party repo checks on ELN
Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-12 16:42:19 -07:00
de36e1179d Revert checkstart usage when launching GNOME Software
We don't have an apps_run_update needle and the later assertions
in this file already make sure Software launches correctly.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-12 16:37:23 -07:00
5f55ba32d7 Remove NetworkManager-config-server after ELN install
This package disables NetworkManager automatic connection creation,
but we really want it.

We *could* write out an explicit config instead, but then we have
to detect the interface name and that's a drag...

We *could* add a test which skips this 'fix' and asserts there's
no network connection on boot, I guess, if we want to test that.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-12 16:33:16 -07:00
e1ba8adf75 g-i-s: don't expect Software screen on ELN
Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-12 16:18:08 -07:00
bce44254a4 Add ELN variant of a gnome-initial-setup needle
The welcome screen text display is slightly different, not sure
why (maybe bigger as it fits on one line?)

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-12 16:08:37 -07:00
0cceba4a4c Add an apps_run_settings needle
Lukas changed a test to use checkstart after launching 'settings',
so we need an apps_run_settings needle.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-12 16:07:08 -07:00
874afba835 Go back to asserting the pgrep check, but give it longer
Seems like GNOME is just having a bit of trouble shutting down on
ELN and it takes longer. So we'll just give it longer.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-12 15:50:36 -07:00
a96b984ab0 g-i-s: extend timeout for pgrep check
Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-12 15:36:38 -07:00
d18ecbc6c9 gnome_initial_setup: try to continue even if pgrep check fails
It's failing on ELN because all the processes stick around as
zombies, for some reason. I am disinclined to look at the reason
right now, so let's do this: even if the check fails, we'll go
ahead and try the userdel. Worst case it fails, and we're no
worse off than before.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-12 14:58:55 -07:00
5a111c1d43 desktop_notifications: add one more gsettings timestamp to be reset
All checks were successful
CI via Tox and perl / tox (pull_request) Successful in 1m18s
CI via Tox and perl / perl (pull_request) Successful in 3m10s
CI via Tox and perl / checkwiki (pull_request) Successful in 32s
gnome-software will look at this key as well, to determine when to pop up the
notification.

Related: quality/tickets#899
Signed-off-by: Kamil Páral <kparal@redhat.com>
2026-05-12 13:54:11 +02:00
0d520ed22d Add a new Maps zoomed-in needle
I think this goes with the recent change to zoom by clicking rather
than shortcut keys, lruzicka forgot to commit it.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-11 17:49:49 -07:00
dffafc743d Add a wrong-font variant webUI language page needle
Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-11 17:48:35 -07:00
f58ee7d86d Add a couple of variant needles for ELN changes
Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-11 17:38:52 -07:00
8e951077f0 Maps: tweak the john o' groats search string
Capitalizing isn't necessary (and is fragile), adding a space
between the ' and the groats gives us better results (i.e. the
first result is what we're looking for, rather than something
weird in America). We need the new needle because previously the
entry we wanted wasn't the first result, so its background color
was different.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-11 17:35:42 -07:00
e2599185d3 Drop abrt handling in privacy page entirely
All checks were successful
CI via Tox and perl / tox (pull_request) Successful in 1m22s
CI via Tox and perl / perl (pull_request) Successful in 3m35s
CI via Tox and perl / checkwiki (pull_request) Successful in 30s
abrt has been removed from Workstation now, so let's just not
expect it here at all.

Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-11 18:30:46 -04:00
7055956cbc Review: implement requests from a review 2026-05-11 18:30:46 -04:00
1a602ac680 optimize required_screens definition to avoid tidy weirdness
Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-11 18:30:46 -04:00
80d173007b Disable DNF check on Silverblue 2026-05-11 18:30:46 -04:00
5517772d9d Pass on Silverblue 2026-05-11 18:30:46 -04:00
2e553bd97e Skip third party on Silverblue 2026-05-11 18:30:46 -04:00
f51054fa72 Set up variable correctly 2026-05-11 18:30:46 -04:00
196a5ba82b Fix langselect 2026-05-11 18:30:46 -04:00
9d33ec211b Add needles 2026-05-11 18:30:46 -04:00
842e9d30d9 Check true for reporting 2026-05-11 18:30:46 -04:00
2e4bc9c68f Check for terminated processes properly 2026-05-11 18:30:46 -04:00
63d2b8432e Add screen checking 2026-05-11 18:30:46 -04:00
7ec2aae187 Fix evaluation 2026-05-11 18:30:46 -04:00
29ceb3f09c Remove milestone flag 2026-05-11 18:30:46 -04:00
68cd324685 aaa_setup: switch back to console if loginctl dumps us at GDM
Signed-off-by: Adam Williamson <awilliam@redhat.com>
2026-05-11 18:30:46 -04:00
469bbe1855 Add new needle to disable reporting 2026-05-11 18:30:46 -04:00
dc587f1df3 Disable automatic reporting during the initial setup 2026-05-11 18:30:46 -04:00
752f345849 Add comment to explain the third_party_repo check 2026-05-11 18:30:46 -04:00
07434abfde Remove workaround for third-party repositories check 2026-05-11 18:30:46 -04:00
d06c713560 Use handle_welcome_screen to clean the Tour 2026-05-11 18:30:46 -04:00
82934b7d76 Rename post_scriptum.pm to gnome_initial_setup_post.pm 2026-05-11 18:30:46 -04:00