Re: [Fwd: Debian specific Partition problme]
I'm guessing it's that the boot disks from Debian are from the
kernel 2.0.x series. It may report different values to fdisk.
Actually, it seems that the debian 2.1 fdisk (kernel 2.0.x) are
reporting physical disk geometry, while the other distros
(probably using kernel 2.2.x series) report translated (lba)
geometry.
Now, the SuSe and RedHat setups may have the same problems, you
just don't know it (yet). You see, if I install debian, and the
kernel is stored in the first 1024 cylinders, all will be well.
however, if I install lots of stuff, and I put the kernel on the
disk, it may end up past the 1024 cylinder mark. The same for
SuSe and RedHat. It just seems that you got lucky when installing
The other two distributions, in that they put the kernel towards
the front of the disk. If you install a lot of stuff on any
distribution, and you install a new kernel, you run the risk of
the kernel file being stored outside the 1024 cylinder range.
this is true of any distribution, and any kernel version.
It is a limit of LILO. Once LILO "starts" the kernel, you should
have no problem. Execpt that LILO can't see the kernel, since
it's so far out on the disk. You have a chicken-and-egg problem.
In short, the fact that fdisk reports differnent values for drive
geometry, and the fact that LILO can't boot your kernel seem to be
two differnet problems. The fdisk geometry problem isn't really
a problem though.
As for a solution, you should do this (regardless of distribution):
If installing Linux on a disk larger that 540Mb, you run the risk
of putting the kernel outside the "reach" of LILO. The best
(IMNSHO) solution is to put a small (20Mb) partition at the
BEGINNING of the disk. It should be type "Linux Native". RedHat
and Debian store the kernel in the /boot directory. The reason
for this is so you can mount the 20Mb partition as /boot. Your
kernel will always be in boot, so it will always be in the first
few cylinders on the disk.
Hope this helps,
Bryan
>
> -------- Original Message --------
> Subject: Debian specific Partition problme
> Date: Tue, 16 Nov 1999 10:18:06 +0530 (IST)
> From: "Nagarjuna G." <nagarjun@hbcse.tifr.res.in>
> To: srivasta@debian.org
> CC: Sudhakar Chandrasekharan <thaths@netscape.com>
>
>
> Dear Manoj/Sudhakar
> I have posted this query earlier on the linux-india list, I found no
> replies.
> I am very curious to know the reason for this behaviour of fdisk program.
> Could you find out this from your friends at debian.org about it?
> the problme is:
>
> I have encountered a peculiar problem on a debian (2.1). The system
> has a 9.1 UltraWide SCSI disk (seagate make). When I do `fdisk' it
> shows the following:
>
> studio:/two# fdisk /dev/sdb
>
> The number of cylinders for this disk is set to 8683.
> There is nothing wrong with that, but this is larger than 1024,
> and could in certain setups cause problems with:
> 1) software that runs at boot time (e.g., LILO)
> 2) booting and partitioning software from other OSs
> (e.g., DOS FDISK, OS/2 FDISK)
>
> Command (m for help): p
>
> Disk /dev/sdb: 64 heads, 32 sectors, 8683 cylinders
> Units = cylinders of 2048 * 512 bytes
>
> Device Boot Start End Blocks Id System
> /dev/sdb1 8 8676 8875912+ f Win95 Extended (LBA)
> Partition 1 does not end on cylinder boundary:
> phys=(1023, 254, 63) should be (1023, 63, 32)
> /dev/sdb5 8 965 979933+ 83 Linux native
> /dev/sdb6 965 1083 120456 82 Linux swap
> /dev/sdb7 1083 4903 3911796 83 Linux native
> /dev/sdb8 4903 8676 3863601 83 Linux native
>
>
> As you notice the number of cylinders are shown to be 8693, too large to
> make lilo work. Obviously I failed to make the machine boot linux, except
> through a boot floppy. Otherwise there are no problems during running.
>
> However, what surprised me was when I tried (on the same harddisk) using
> SuSe 6.2, and Redhat 6.0, the out put of fdisk shows entirely different.
> They show only 1106 cylinders. This time lilo also works fine.
>
> I thought the fdisk program is same across different distros.
> Why does this happen? Any clues?
>
> Nagarjuna
>
>
>
>
> ------------------------------------------------------------------------------
> Nagarjuna G.
> Homi Bhabha Centre for Science Education, TIFR, Mumbai. INDIA
> Phones: Office: 556 7711, 555 4712, 555 5242 Residence: 2155604
> Fax: 091 - 22 - 556 6803
> email: nagarjun@hbcse.tifr.res.in
> ------------------------------------------------------------------------------
>
>
> --
> Unsubscribe? mail -s unsubscribe debian-user-request@lists.debian.org < /dev/null
>
>
Reply to: