This broke with Fedora-27-20171012.n.0, when GNOME 3.26.1 landed in Fedora 27. When you run a Workstation install without creating a user account in the installer, on boot, gnome-initial-setup should run in its 'user account creation' mode, allowing (in fact requiring) you to create a user account. However, it no longer does. On first boot after a Workstation install without a user account created, the system boots to GDM, displaying a 'Username:' prompt: https://openqa.fedoraproject.org/tests/158599#step/_graphical_wait_login/8 This doesn't actually appear to violate the release criteria as written, but it is quite a bad bug, so nominating for a Final freeze exception.
so i presume this is caused by this commit: https://git.gnome.org/browse/gdm/commit/?id=1abdb kernel_cmdline_initial_setup_argument is supposed to fail unless gnome.initial-setup is on the kernel command line, but I guess it must be succeeding? I'm guessing openqa doesn't run with gdm debug enabled right? I'll probably just need to fire off an install, unless someone else already has journal with gdm debug enabled handy.
oh i bet i know, i bet it's trying wayland and failing for whatever reason, and then falling back to X, but now already_done_initial_setup_on_this_boot succeeds.
Ray, I think Pop!_OS had a similar problem as this bug (Initial Setup's new user mode not working with Wayland) and it was one reason they disabled Wayland for their 17.10 release.
Discussed during the 2017-10-23 blocker review meeting: [1] The decision to classify this bug as an AcceptedFreezeException was made as this cannot be fixed with an update. [1] https://meetbot.fedoraproject.org/fedora-blocker-review/2017-10-23/f27-blocker-review.2017-10-23-16.00.txt
Jeremy: Ray's theory is actually the opposite: that it works if Wayland does (or if the system is configured such that no attempt is made to run g-i-s under Wayland at all), but doesn't work if we hit a fallback from Wayland to X11, because the 'OK, we ran already' file gets written on the failed attempt to run it under Wayland, so when it's run again under X11, it decides it ran already and quits. As I understand his theory, anyhow.
Ray posted a scratch build: https://koji.fedoraproject.org/koji/taskinfo?taskID=22653424 in my testing, it looks like it fixes the bug. thanks.
*** Bug 1505997 has been marked as a duplicate of this bug. ***
This also affects the armhfp and aarch64 disk images, booting to gdm rather than gnome-initial-setup. Confirmed fixed with gdm-3.26.1-2.fc27.
Is it going into Bodhi?
Any test Live images available with this build yet?
No, but you don't really need one. Just run an install from a recent live image, then do 'chroot /mnt/sysimage' and update the package after the install completes, before rebooting to the installed system. Ray, can you submit an update with the fix? Thanks!
╎rstrode@halfstation〉⎛/srv/sources/pkgs/gdm⎞ ⌜03:46 PM⌟ ╎❯ fedpkg update Password: ================================================================================ gdm-3.26.1-2.fc27 ================================================================================ Update ID: FEDORA-2017-2a7d1f7ebf Content Type: rpm Release: Fedora 27 Status: pending Type: bugfix Karma: 0 Autokarma: True [-3, 3] Request: testing Bugs: 1502827 - None Notes: make sure initial-setup starts when wayland fails Submitter: rstrode Submitted: 2017-10-26 16:57:38 Comments: bodhi - 2017-10-26 16:57:38 (karma 0) This update has been submitted for testing by rstrode. https://bodhi.fedoraproject.org/updates/FEDORA-2017-2a7d1f7ebf
gdm-3.26.1-2.fc27 has been pushed to the Fedora 27 stable repository. If problems still persist, please make note of it in this bug report.