Re: What happened to debian - does "stable" keep having any meaning?

[Tom H <tomh0665@gmail.com>]

On Wed, Mar 23, 2011 at 1:59 AM, Geronimo <geronimo013@arcor.de> wrote:
> Boyd Stephen Smith Jr. wrote:
>>
>> Now, moving between 6.0.0 and 6.0.1 shouldn't have been a problem, but I
>> suspect you actually would have had issues rebooting your 6.0.0 system even
>> without the 6.0.1 updates, since you didn't have your fstab in order.
>
> Grub can't be wrong, cause it is working in your system.
> So, the only conclusion is: my fstab must be wrong.
>
> I think, that's the same wrong reduction of facts, than my saying: grub is
> wrong.
>
> So let me state, debian 6.0 was working fine til the last update.
>
> On a complex system (yes, for me, linux is quite a complex system) the truth
> may be somewhat different. May be, we both are wrong.
>
> So I did some researches with my system constellation (all 8 drives active):
>
> 1.) A fresh installation from debian 6.0 netinst CD results in an unbootable
> system, even using a single partition installation target.
>
> 2.) following the advices of:
>        grub-mkdevicemap
>        grub-install /dev/disk/by-id/ata-INTEL_SSDSA2M040G2GC_CVGB0061021D040GGN
>        update-grub
>  results in an unbootable system
>
> 3.) at first sight it looks like ubuntu 10.10 is using the same grub variant,
> but a closer look shows, that on ubuntu there's no /boot/device.map
>
> 4.) booting the debian 6.0 netinst CD in rescue64 mode and chrooting the
> ubuntu installation, a "grub-install /dev/sde" brings my system back into play
>
> Well, I stil believe, that some of the last update was not good enuf for
> debian stable ...

- grub wasn't updated for 6.0.1. Its last squeeze update was on 17
January, pre-6.0 release (6 February IIRC). So, strictly speaking, it
isn't grub that's broken grub.

- I've never understood the device.map difference between Debian and
Ubuntu because the maintainers seem to be the same. That file's only
used at install time so, perhaps, it's created and deleted on Ubuntu
in some kind of (strange) cleanup effort.

-  Do the grub.cfg entries for squeeze in squeeze and maverick match
for the insmods, the grub root, and the system root?

- Does grub-probe on squeeze and maverick return the same values?

On squeeze, for example with:

for directory in / /boot; do echo "### PROBE OF $directory ###"; for
target in fs fs_uuid device drive abstraction; do /usr/sbin/grub-probe
--target $target $directory; done ; done

for systemdevice in /dev/root_partition /dev/boot_partition; do echo
"### PROBE OF $systemdevice ###"; for target in fs fs_uuid device
drive abstraction; do /usr/sbin/grub-probe --target $target --device
$systemdevice; done ; done

On maverick, for example with:

for directory in /mnt/squeeze /mnt/squeeze/boot; do echo "### PROBE OF
$directory ###"; for target in fs fs_uuid device drive abstraction; do
/usr/sbin/grub-probe --target $target $directory; done ; done

for systemdevice in /dev/squeeze_root_partition
/dev/squeeze_boot_partition; do echo "### PROBE OF $systemdevice ###";
for target in fs fs_uuid device drive abstraction; do
/usr/sbin/grub-probe --target $target --device $systemdevice; done ;
done

- Can you chainload squeeze's grub from maverick's?

I haven't set up a chainload or used 40_custom in a *long* time but I
think that the following should create the right menu entry.

$ vi /boot/grub.d/40_custom
#!/bin/sh
cat << EOF
menuentry "chainload squeeze from maverick" {
set root=(hdX,Y)  ## using X and Y for /mnt/squeeze/boot
linux /grub/core.img ## or /boot/grub/core.img if / and /boot aren't
different partitions
}
EOF
$ update-grub