On Mon, Jul 13, 2015 at 11:33 AM, John Wright <jwright2@xxxxxxxxxx> wrote: > > On Mon Jul 13 02:19:31 UTC 2015 Chris Murphy typed: > >> So there's the kernel bug, 1225671, that ends up stopping the arrays, >> but then there's a misleading message saying there's a problem that's >> been corrected, yet clearly not corrected. > > > The exact wording of the message is: > > 'Unexpected system error. The system has encountered a problem and > recovered.' This is a message in the installer itself? Or is this a GNOME notification (a floating thing from the top-center)? I'm going to guess it's not in the installer because I've always seen it spit out the exact error message into the anaconda.log (or program.log if it's a helper program's error) and I don't see that in your attached logs. But I also don't see it in the journal either, so... I'm baffled exactly what system error it's referring to as well as this supposed recovery. GNOME bug... haha. > >> Does this same kernel call trace happen with Fedora 21's installer? Or >> does it fail for a different reason? > > > With the F21 installer, there is no error message, and the installer > does find the raid1 array. But it finds it only as one ~ 460 GB > 'physical volume'; my existing LVM partitions are not recognized, so > I can't install into them. Goofy cakes. OK so that suggests PV metadata is recognized, but for some reason the VG/LV's are actually activated for some reason. This could be a variation of bug 825236, which causes linux installs on LVM to be recognized by grub-mkconfig; but you don't have the problem on Fedora 20 which is likewise afflicted. At least the VG should be available so you can add (space permitting) new LVs. But if the LV's aren't made active and listed, you couldn't shrink any of them to make room to make additional ones. Blek. So as a work around for Fedora 22's installer, I wonder if this oops is fatal or if you can assemble the array with mdadm -A before launching the installer, get the array going again, then lvchange -ay to activate the LVs, and then launch the installer. Either it'll see the LV's now, *OR* you get another kernel oops and the array stops again making everything go away. If the former, work around seems likely. If the latter, then you'd need to build your own installer ISO swapping out the kernel version for 4.0.5 or greater which should fix the problem. > Alas, rolling my own workstation installer is beyond my present skill > level. I must decide whether to wait for Fedora 23, or to back up > everything and let the F22 installer have a crack at creating a new > raid1 array with LVMs on top of it. Seeing as this is broken in two different ways with Fedora 21 and 22, it would benefit the project, other users, and yourself, to test this in Fedora 23 at some point. And if it's still not working, file a new bug if the failure doesn't match up with one of the existing bug reports, and mark as a blocker. More info on test@ list how to go about doing this. -- Chris Murphy -- users mailing list users@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines Have a question? Ask away: http://ask.fedoraproject.org
