[Tfug] Noobie Problems with Upgrade to Sarge

[Stephen Hooper]

On Tue, 2005-06-28 at 23:34 -0400, Charles R Kiss wrote:

> 
> *house:/boot# lilo*
> Warning: /dev/hdb is not on the first disk
> Added Linux *
> Added WindowsXP
> Syntax error at or above line 13 in file '/etc/lilo.conf'
> 

Looks like you are telling LILO to install on the MBR of the second IDE
disk (hdb), which is probably not what you want... 

> 
> *lilo.conf:*
> 
> *charles at house:~$ cat /etc/lilo.conf*
> boot=/dev/hdb
> root=/dev/hdb5
> prompt
> install=/boot/boot-menu.b
> map=/boot/map
> vga=normal
> delay=20
> image=/vmlinuz
> label = Linux
> read-write
> other=/dev/hda1
> label = WindowsXP
> initrd=/initrd.img
> 

That seems to confirm it :}

So maybe change the "boot" line to /dev/hda (primary IDE MBR).

I don't see 13 lines in what you pasted.

I would also take the time to normalize the file (remove strange
characters), and remove the spaces from around the "=".  That is pure
guesswork on my part, as my machine is now grubby, and I don't have the
lilo man page memorized :(

Not sure, if Debian does this, but you too seem to have a "/boot"
directory.  I would always keep my initrd, and kernel in there.  I guess
that may just be the RedHat/Gentoo in me.

It does make for a nice clean "/" though, and has advantages in that
once booted you can unmount the "/boot" fs (if you made it a different
fs), and save yourself possible corruption of the filesystem if the
machine were to go down hard (so you should always at least be able to
boot the kernel without digging around for floppies).

I guess that with LILO it doesn't make much difference because LILO
throws the kernel in with the boot loader stuff somewhere (maybe in the
MBR).

> *here is fdisk -l*
> 
> *house:/home/charles# fdisk -l*
> 
> Disk /dev/hda: 60.0 GB, 60022480896 bytes
> 255 heads, 63 sectors/track, 7297 cylinders
> Units = cylinders of 16065 * 512 = 8225280 bytes
> 
>    Device Boot      Start         End      Blocks   Id  System
> /dev/hda1   *           1        2550    20482843+   7  HPFS/NTFS
> /dev/hda2            2551        7296    38122245    f  W95 Ext'd (LBA)
> /dev/hda5            2551        7296    38122213+   7  HPFS/NTFS
> 
> Disk /dev/hdb: 137.4 GB, 137438952960 bytes
> 255 heads, 63 sectors/track, 16709 cylinders
> Units = cylinders of 16065 * 512 = 8225280 bytes
> 
>    Device Boot      Start         End      Blocks   Id  System
> /dev/hdb1   *           1           1        8001   83  Linux
> /dev/hdb2               2       16709   134207010    5  Extended
> /dev/hdb5               2        7295    58589023+  83  Linux
> /dev/hdb6            7296       14590    58597056    6  FAT16
> /dev/hdb7           14591       15441     6835626   82  Linux swap / Solaris
> 
> */dev/hda is my WindowsXP partition, which /IS/ the first disk.

> here is uname (because I'm booting from the floppy):
> *Linux house 2.4.18-bf2.4 #1 Son Apr 14 09:53:28 CEST 2002 i686 GNU/Linux*
> 
> Here are the files contained in / (minus the directories):
> *initrd.img     
> initrd.img.old
> vmlinuz*
> *vmlinuz.old

> *I'm still learning about the new initrd process; I guess it boots off 
> RAM which is loaded from somewhere.  I'll figure that out.
> 

initrd's are small ramdisks that get run as a "/", before the "real"
root is run.  The only times I have ever needed it are when I load
drivers for a device that must be present at boot time, but for which I
don't want to compile a kernel (SCSI controllers, LVM root disks); or
when you want a fancy console display.  Other people probably use it to
take over the world.

"man initrd" is a good place to start on your quest for understanding.

>  < *snip* >

> 
> The wierd thing is that my desktop is now in GNOME; I never used GNOME, 
> but somehow that's what I get when I boot up with my old floppy.  I 
> kinda like gnome. All the packages are new, including openoffice, but 
> which doesn't open (ofcourse).
> 

That is probably the fault of upgrading all the packages, and not of
booting off your floppy.  Look at the script files in /etc/X11 to
understand what it is doing, but basically:

- Probably looks to see if Gnome GDM, or KDE KDM is installed, and
starts one of those.
- If it can't find either one of those, then probably it probably looks
to see if XDM is installed,  and starts it.
- Whatever above presents you with a login screen, which you then log
into :)

The  session script is then run, which probably:

- Looks to see if you have specified a WM for the session.
- If it doesn't find that, it then looks to see if you have ~/.xsession,
~/.xinitrc.
- If those files don't exist it then probably just looks for Gnome or
KDE.
- If it doesn't find either of those then it looks for other Window
Managers (WM, Fluxbox).
- If it doesn't find those it probably chooses the "Failsafe" option
which is usually twm.

So probably what happened is when you used the "apt" thingy, you pulled
down a Gnome install, and you have no .xession set up.

That is based off of memory from RedHat installs, don't take it to line
up perfectly with your distribution.

The scripts are usually called something like "*chooser*", and something
like "*session*" (or maybe "*dm*", and "*startup*").  I think just
grepping for references to "xdm", and probably "twm" would give you both
scripts of interest (if you are interested in the way X's sessions start
up on your machine).

The floppy should just be using "/dev/hdb1" as the root, which is why I
say it is probably the package installs you just ran, and not the
floppy. 

The X session stuff doesn't usually happen until a much higher level
(which is usually "described" (if you want to call it that) in your
"root" fs ).

Hope you don't mind that I responded even if I am not a Debian user.
You will probably get nicer answers from someone who actually runs
Debian.