Description of problem: I had a partially-intalled copy of F13-beta on my h/d, no bootloader was installed. Upon restarting installation in text mode, I chose 'upgrade'. It prompted me to skip bootloader configuration or install a new one. I chose the latter. The installer then said there was a backtrace in 'bootloader' config. All this is from the f13-beta-netboot.iso.
BTW I initially tried the same thing in the graphical interface and the crash box didn't come up there -- everything had just frozen. Switching VTs I poked around in a few log files and all had old timestamps, so obviously no progress was being made. The crash doesn't happen immediately after the bootloader screen, it proceeds to the next 'active ethernet connection' screen, then says it'll fetch the repo information and this is when the crash happens. Anaconda version is 13.37.2.
Can you please attach the traceback you are getting? You should be able to do it using error report dialog. If there is no dialog/traceback - e.g. just freeze in graphical UI case, you can go to VT2, find command starting "kill --USR2" in bash command history (using up arrow several times), run it, and the traceback will be dropped into /tmp/anaconda-tb* file.
I didn't see any way of getting a useful traceback then. As I mentioned, the GUI had just frozen but the text-mode install gave me an option to get to pdb, which didn't turn up useful info AFAIR. Anyway I don't have that state of the install right now, but it's easily reproducible in the manner I've described. If I do get into such a situation again, I'll post the trace.
(In reply to comment #0) > Description of problem: > > I had a partially-intalled copy of F13-beta on my h/d, no bootloader was > installed. > This condition can be pretty particular and hard to reproduce as "partially-installed" is too vague description. I'd need more specific steps to reproduce the issue, and preferably the traceback to be able to consider whether this is an issue to work on in the first place.
I was installing via the netboot.iso cd and had a network glitch during package installs (it was openoffice.org-core that was being installed at the time). Install seemed to have stalled. After several attempts to find out if things were proceeding (by checking for logs, downloaded rpms, etc.), I gave up and hit 'reboot' on VT2. I then proceeded to do an 'upgrade' via GUI, which got stuck again, so I re-tried upgrade using the text-mode install. After that was done, I installed the remaining packages (kernel and grub, importantly) to get to a working system to continue from there.
If you can reproduce, you can jump over to tty2 and scp /tmp/anaconda-tb-* off the system to attach to this bug report.
Well I just cloned an F11 VM, ran dd if=/dev/zero of=/dev/sda1 bs=512 count=1 and ran the installer in the text mode, got the traceback. Attaching the dumps.
Created attachment 405460 [details] Anaconda dump 1
Created attachment 405461 [details] Anaconda dump 2
BTW the GUI install had frozen in the VM as well, with no idication of a crash.
anaconda 13.37.2 exception report Traceback (most recent call first): File "/usr/lib/anaconda/text.py", line 597, in run (file, classNames) = stepToClasses[step] File "/usr/bin/anaconda", line 1188, in <module> anaconda.intf.run(anaconda) KeyError: 'bootloader'
Why are you using text mode for this? Also I'm not really sure how valid a bug report this is given that your starting situation is a "partially installed copy of F13".
(In reply to comment #12) > Why are you using text mode for this? Because the graphical mode just freezes and doesn't even tell me there was a crash. I've mentioned this above. > Also I'm not really sure how valid a bug > report this is given that your starting situation is a "partially installed > copy of F13". I've also mentioned that this happens when I upgrade from F11 with grub removed.
*** Bug 590823 has been marked as a duplicate of this bug. ***
Tested http://clumens.fedorapeople.org/updates.img and confirm that 'New bootloader configuration' is no longer an option in text-mode upgrades
Given that this is specific to text mode, I do not think it qualifies as a blocker -- especially at this point in the schedule. There are too many other (better) ways to get this done.