Grub Chaining Problem
Gail Koontz
with a grub boot to a small system on sda and chaining from there to
other systems on both sda and sdb. This has always worked just fine. I
encountered a problem when I attempted to add four partitions to the
extended partition on sda: sda15-sda18. All were allocated by diskdrake
with no problem, and sda15 was successfully formatted. However, the
attempt to format sda16 wiped out the entire b drive (yes, really, and
it's repeatable!) No problem to recover the data on it, but since then
I have been unable to chain to any system created on the b drive. I can
set up to boot to it directly but not to chain to it. Can anyone
suggest what might have happened to that drive to prevent this and how
I might correct it? I can't install a new system until I can either
create partitions on the a drive or chain to a system on the b drive.
--
Gail Koontz Retired in my home state
836 Mallard Rd. ... and loving it!
Cocoa, FL 32926 gailk...@cfl.rr.com
David W. Hodgins
> I have two drives, sda and sdb. I have been operating for a long time
> with a grub boot to a small system on sda and chaining from there to
> other systems on both sda and sdb. This has always worked just fine. I
> encountered a problem when I attempted to add four partitions to the
> extended partition on sda: sda15-sda18. All were allocated by diskdrake
> with no problem, and sda15 was successfully formatted. However, the
> attempt to format sda16 wiped out the entire b drive (yes, really, and
sd drives are accessed as scsi drives, whether they are, or not, and are
limited to 15 partitions. It's definitly a bug in diskdrake, that it even
tries to create sda16, let alone claims to have successfully allocated it,
when the allocation has actually failed.
See http://qa.mandriva.com/show_bug.cgi?id=14696
To get around the limit, consider using lvm phsyical volumes, where one
partition on the hd can contain many logical volumes, each with there own
filesystem.
Regards, Dave Hodgins
--
Change nomail.afraid.org to ody.ca to reply by email.
(nomail.afraid.org has been set up specifically for
use in usenet. Feel free to use it yourself.)
Gail Koontz
> See http://qa.mandriva.com/show_bug.cgi?id=14696
I certainly missed that. Thanks!
Do you have any idea what changed on the b drive that prevents my
chaining to a system there?
David W. Hodgins
> David W. Hodgins wrote:
>> See http://qa.mandriva.com/show_bug.cgi?id=14696
> I certainly missed that. Thanks!
> Do you have any idea what changed on the b drive that prevents my
> chaining to a system there?
Either the recovery failed, or the partition numbers have changed.
Post the output of your grub menu.lst, and the output of sfdisk -l
(or fdisk -l). Mount the boot filesystem under something like
/mnt/hdb5 (using the correct partition number of course), and post
the output of ls -l /mnt/hdb5. Also mount the root as something
like /mnt/hdb6, as post the output of cat /mnt/hdb6/etc/fstab.
mark south
>> encountered a problem when I attempted to add four partitions to the
>> extended partition on sda: sda15-sda18. All were allocated by diskdrake
>> with no problem, and sda15 was successfully formatted. However, the
>> attempt to format sda16 wiped out the entire b drive (yes, really, and
>
> sd drives are accessed as scsi drives, whether they are, or not, and are
> limited to 15 partitions. It's definitly a bug in diskdrake, that it even
> tries to create sda16, let alone claims to have successfully allocated it,
> when the allocation has actually failed.
It rather sounds as if something mapped "sda16" to the next drive along,
which would have been sdb....
David W. Hodgins
Unlikely. If you run rpmdrake from a console, when it tries to allocate
sda16, you'll see the error mesages in the console. However, rpmdrake
will indicate the allocation was successfull. Similar with format,
and mount. They fail, but the gui in rpmdrake looks as if it has
succeded.
Gail Koontz
>
> It rather sounds as if something mapped "sda16" to the next drive
> along, which would have been sdb....
Exactly. It is probably the classic case of overflowing a space meant to
hold some maximum and and ending up with 0. Maybe 0f is drive 0,
partition 15 and 10 is drive 1, partition 0 and succeeds in wiping out
either the whole drive or the partition table, which amounts to the
same thing if you don't try restoring the partition table, which I
didn't. Diskdrake will show the illegal partition 16 on a as allocated
and happily wipe out the b drive on any attempt to format partition 16.
Ron Gibson
>> It rather sounds as if something mapped "sda16" to the next drive along,
>> which would have been sdb....
> Exactly. It is probably the classic case of overflowing a space meant to
> hold some maximum and and ending up with 0. Maybe 0f is drive 0, partition
> 15 and 10 is drive 1, partition 0 and succeeds in wiping out either the
> whole drive or the partition table, which amounts to the same thing if you
> don't try restoring the partition table, which I didn't. Diskdrake will
> show the illegal partition 16 on a as allocated and happily wipe out the b
> drive on any attempt to format partition 16.
Sometimes the default is to only make 16 partions per drive, a headache
I run into at times. (IOW /dev/hda15 is the last one made for hda).
Is this what you ran into? Check /dev just in case.
If that's the problem there is an easy fix.
Yeah I've booted a live CD before to rescue a partition and when trying
to mount /dev/hda18 I get something like "Device not found" - That's
always a good one for a scare...
--
Linux Help: http://rsgibson.com/linux.htm
Email - rsgi...@verizon.borg
Replace borg with net