On Sat, 15 Sep 2012 02:19:11 +0000, Hendrik Boom wrote: > On Fri, 14 Sep 2012 22:37:28 +0100, Joe wrote: > >> On Fri, 14 Sep 2012 20:42:21 +0000 (UTC) >> Hendrik Boom <hend...@topoi.pooq.com> wrote: >> >>> On Fri, 14 Sep 2012 23:25:03 +0300, Andrei POPESCU wrote: >>> >>> > On Vi, 14 sep 12, 17:12:38, Hendrik Boom wrote: >>> >> >>> >> Of course, after I've made my copy (with slight changes to >>> >> /etc/fstab) I have two nearly identical sets of partitions, so it >>> >> may be tricky to tell them apart. Is grub2 clever enough to figure >>> >> it all out anyway? And what data does it use to this end? (so I can >>> >> make sure it's right!) >>> > >>> > UUIDs? What failure mode(s) do you have in mind, because I can't >>> > think of any. >>> >>> It probably is os-prober that I mean. The misconfiguration I have in >>> mind is matching one system's /boot with another systems's /. I've >>> had it happen on a laptop sometime ago. and it sure messed up my >>> upgrades. I have no idea how it happened, but it has made me paranoid. >>> >>> >> The problem is that update-grub rewrites /boot/grub/grub.cfg. It may be >> possible to specify roots and boots in /etc/grub.d/ (I do use a >> separate /boot, but I've never needed to try this) or alternatively it >> is perfectly possible to edit grub.cfg, but you need to remember to do >> so each time update-grub is run, before rebooting. More than once, I've >> known versions of grub not deal correctly with a separate /boot, so >> I've had to do this until the bug was fixed. > > It might be that bug that fouled up my laptop. Or I might have done it > myself at some point. Anyway, I've become paranoid about such things. > If the bug has been fixed, well, that'll be one less thing to go wrong > when I upgrade. > > Thanks for telling me that there was a bug, and that it was fixed.
Actually, it seems it may be bug #612407 -- the symptoms are almost exactly what happened on my laptop, with the exception that I had the problem with sqeeze/wheezy instead of lenny/squeeze: http:// bugs.debian.org/cgi-bin/bugreport.cgi?bug=612407 If they've fixed the bug as you say, this is presumably another one, unless they've fixed it but failed to mark it as fixed. In any case, this tells me that I should go ahead with my upgrade plans, but should watch any automatically updated grub configuration files like a hawk, diffing them against previous versions, and the like. Also, looking around the grub buglog, it looks as if booting several different / partitions from one /boot partition is perfectly legal. But I don't know if the Debian package manager is cool with that -- one partition being upgraded independently by two package managers? It seems they might disagree about when files need to be removed. It's copnceivable that I made a mistake on my laptop with /etc/fstab and inadvertently did specify the same /boot for the different systems. Or that it happened during the automatic upgrade of /fstabs from /dev/sd* notation to UUID notation. But the number of comments I've seen about os- prober getting things wrong makes me suspect that the trouble could arise even without a faulty /etc/fstab. I just don't know any more. What was it the Defence Against the Dark Arts master kept saying all the time? Something like "Extreme Vigilance"? At least I know what to beware of now. -- hendrik -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/k32na0$q76$4...@ger.gmane.org
