Bug#432650: closed by maximilian attems <email@example.com> (Re: Bug#432650: Sigh...)
Well, the initial information was likely insufficient. I
have more data points now:
I am running Debian Lenny on an ASUS L2D laptop. Up until
kernel 2.6.21 got installed by apt the primary disk was
detected as an IDE disk (hda) and interfaced as such.
Starting with kernel 2.6.21 it got detected as an SCSI disk
(sda) and gets interfaced via SCSI-controlled-ATA. At that
moment the system stopped booting (via 2.6.21) since it
couldn't find the root partition anymore.
Once I manually fixed
- /boot/grub/menu.lst (root device, resume device, run update-initramfs/update-grub)
- /etc/fstab (swap and other partitions)
things now seem back to normal (of course, now I cannot boot
the old 2.6.18 kernel anymore since it requires /dev/hda
instead of /dev/sda).
What I consider a bug is that the upgrade didn't tell me
about the chance of partitions being differently detected
now and not pointing out that I may need to adjust a bunch
of config values.
It would be very nice - but not mandatory - for the upgrade
process to ask and go change the configuration itself if
needed. Which may be against the policy of one package not
changing the configuration of another one. Which could be
overcome by doing dpkg-reconfigure <package> on the packages
So, all in all this bug isn't really closed. It is certainly
not a kernel bug but rather an annoyance in the way the
upgrade is handled. I understand it may be difficult to know
just when a kernel upgrade will "break" on a particular
machine. If that's so the *information* put out during
kernel upgrades should be improved. I hope I didn't just
miss that message. I usually read all of them.
GPG key ID E4071346 @ wwwkeys.pgp.net
E167 67FD A291 2BEA 73BD 4537 78B9 A9F9 E407 1346