Created attachment 2111072 [details] /tmp/journal.log Installer WebUI Critical Error: Connection has timed out. Please attach the log file /tmp/journal.log to the issue. ---[ System & Environment Information ]--- OS: Fedora Linux 43 (Workstation Edition) Anaconda version: 43.44 Anaconda UI version: 53.14.g7ea927aa
Created attachment 2111073 [details] /tmp/anaconda-webui.log
This was a reinstall of F43 on a machine where I had already done a F43 install which failed near the end. I tried the trick of deleting the /var/lib/machines subvolume before installation but it didn't help. After the last install, I at least had a 2 GiB /boot and the machine seemed perfectly functional so hopefully I won't be any worse off this time. I've had the same type of trouble on 3 of my machines, the only thing they have in common is that they're old and slow. The only two fully successful installs were on the newest machine which is a Lenovo X131e from 2013, and even on that machine, most attempts failed.
Also see my comments 13-16 in https://bugzilla.redhat.com/show_bug.cgi?id=2406198 . I initially thought this was related but the fix for that (deleting /var/lib/machines before install) doesn't work here. I haven't noticed any issues with the mostly updated machines, though. I'm mostly concerned with data corruption so would like to know if that's possible or what other issues might arise.
Should note that in most of my previous attempts, I never got an explicit crash, I just noted that nothing was happening (except for the progress circle continuing to spin, but no significant processes were running), gave up and rebooted into an apparently working system. I eventually figured out that I could trigger a crash by clicking on the Storage Editor button in the upper right, that wakes it up enough to realize that it's failed.
Are there other nested volumes? It's not common but some container managers use btrfs subvolumes. # btrfs sub list /path/to/mountpoint >I'm mostly concerned with data corruption so would like to know if that's possible or what other issues might arise. Low likelihood of data corruption. Potential for out of order writes if the drive is not consistently honoring FLUSH or FUA commands, AND if there's a crash or power fail. This tends to result in boot failure following the crash, and we have all kinds of bad UX for this scenario. Best to discuss offline.
(In reply to Chris Murphy from comment #5) > Are there other nested volumes? It's not common but some container managers > use btrfs subvolumes. > > # btrfs sub list /path/to/mountpoint I know almost nothing about btrfs so wouldn't know how to create additional volumes, I'm just using what's generated by a clean install in a dual-boot system with Windows on sda1 and sda2 (except I deleted /var/lib/machines before the latest attempt associated with this bug). I saw nothing else nested when I used the Storage Editor. And a quick check on the machine I'm currently using, which has identical partitioning, shows nothing other than root/var/lib/machines which I deleted on the other one.