In Rawhide (and occasionally F44) openQA tests recently, we frequently see Workstation (but not KDE) live install tests fail in a distinctive way. The installer UI gets close to the end of the process - usually showing "Configuring installed system", with a big spinner at the top - and just gets stuck like that, for over an hour. However, the journal indicates the install actually completed, shortly after the UI reached that state, e.g.: Feb 13 12:56:12 localhost-live org.fedoraproject.Anaconda.Boss[3418]: INFO:anaconda.modules.boss.installation:All tasks in the installation queue are done. Installation successfully finished. Feb 13 12:56:12 localhost-live org.fedoraproject.Anaconda.Boss[3418]: INFO:anaconda.core.threads:Thread Done: AnaTaskThread-RunInstallationTask-1 (281473424748928) So it seems like the installer UI (which is running in Firefox) gets stuck somehow. This never seems to happen on KDE (where the installer UI runs in slitherer). It seems like it's happening more often on Rawhide than on F44, but that *could* just be chance. It happens on both x86_64 and aarch64. Filing against Firefox on the tentative theory this is a bug in FF. The earliest case of this I can find is https://openqa.stg.fedoraproject.org/tests/5894115 , which is from 2026-02-10, between firefox-147.0.3-1.fc45 (Feb 6) and firefox-147.0.4-1.fc45 (Feb 14).
This sounds exactly the same as bug 2442147.
As I mentioned in the other bug, the difference is that in this bug, the log clearly says that the install finished, but in my bug, it doesn't (which makes it more serious). I don't know if that's because the log in my case needs to be "flushed" and isn't showing everything that actually happened.
Nominating for Freeze Exception since many people won't know how to check the logs to see that the install actually finished. In https://bugzilla.redhat.com/show_bug.cgi?id=2442147 it's more serious since the logs don't even show that.
I don't think this and 2442147 are quite the same, for the reasons Andre noted. We definitely only started seeing this in openQA relatively recently, we did not see it on F43.
Also, checking back in, the pattern that this is much more common on Rawhide than F44 seems to have persisted. I've no idea why. If we're theorizing it's to do with RAM exhaustion, though, I guess RAM usage could be higher on Rawhide for some reason. We should probably use the anaconda feature to profile it.
AGREED AcceptedFinalFreezeException Discussed at the 2026-03-30 (blocker / freeze exception) review meeting: There seem to be various mysterious aspects to this, but we agree that if somebody somehow figures out what's going on and finds a safe fix, the issue certainly qualifies as a Freeze Exception in principle. It's a very visible and apparently-critical issue that cannot be fixed with an update, even if in fact the install probably went fine and will work on reboot. https://meetbot-raw.fedoraproject.org//blocker-review_matrix_fedoraproject-org/2026-03-30/f44-blocker-review.2026-03-30-16.00.log.txt
Conversely, it's possible to get a screen showing a successful install, but the logs in /tmp are truncated and don't show that the install completed (in this case I'm assuming it actually did). See https://bugzilla.redhat.com/show_bug.cgi?id=2442147#c37 .
Common Issue draft: https://discussion.fedoraproject.org/t/common-issue/189635
I don't actually recall seeing this one on openQA much lately, it's possible it went away somehow? I'll have to check some test result history.
oh, nope, hey, here it is, happening again: https://openqa.fedoraproject.org/tests/4658413#step/_do_install_and_reboot/106
Adam, has it every happened for anything else than Workstation Live? I'm mainly interested if you've seen this in the anaconda GTK interface, because that can be one of the proposed workarounds for users affected by this.
I've never seen it with the GTK interface. The problem started for me exactly when WebUI did. Unfortunately, as I recall, when installing from the F43 Live I found that attempting to remove anaconda-webui (to get the GTK interface) would remove anaconda packages as well so I could no longer run the Live installer that way. And prior to this problem, I never checked if the logs in /tmp were truncated, although I now know that can happen even when getting the graphical "install successful" screen. I'll be checking every time from now on.
I don't think I've seen it with gtkui.
Thanks, I've published the common issue description citing anaconda GTK as a suggested workaround.