[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Installing Debian on a Qube 1



An initial mixed success ... the machine hung at the stage where the LCD
read "Booting Debian" (i.e. after colo runs)

syslog suggests that this is hanging on setup of eth0 ... oddly, it also
says that there are two nics on the machine (there's only one external
port on a Qube 2700, and no sign of another interface ...

Nov 13 05:53:10 localhost kernel: Linux Tulip driver version
0.9.15-pre12 (Aug 9, 2002)
Nov 13 05:53:10 localhost kernel: PCI: Enabling device 00:07.0 (0045 ->
0047)
Nov 13 05:53:10 localhost kernel: tulip0: Old format EEPROM on 'Cobalt
Microserver' board.  Using substitute media control info.
Nov 13 05:53:10 localhost kernel: tulip0:  EEPROM default media type
Autosense.
Nov 13 05:53:10 localhost kernel: tulip0:  Index #0 - Media MII (#11)
described by a 21142 MII PHY (3) block.
Nov 13 05:53:10 localhost kernel: tulip0:  MII transceiver #1 config
1000 status 7809 advertising 01e1.
Nov 13 05:53:10 localhost kernel: eth0: Digital DS21143 Tulip rev 65 at
0x100000, 00:10:E0:00:BE:5E, IRQ 4.
Nov 13 05:53:10 localhost kernel: PCI: Enabling device 00:0c.0 (0005 ->
0007)
Nov 13 05:53:10 localhost kernel: tulip1: Old format EEPROM on 'Cobalt
Microserver' board.  Using substitute media control info.
Nov 13 05:53:10 localhost kernel: tulip1:  EEPROM default media type
Autosense.
Nov 13 05:53:10 localhost kernel: tulip1:  Index #0 - Media MII (#11)
described by a 21142 MII PHY (3) block.
Nov 13 05:53:10 localhost kernel: tulip1:  MII transceiver #1 config
1000 status 7809 advertising 01e1.
Nov 13 05:53:10 localhost kernel: eth1: Digital DS21143 Tulip rev 65 at
0x101000, 00:10:E0:00:C5:7A, IRQ 13.
...
Nov 13 05:53:10 localhost kernel: eth0: Setting full-duplex based on
MII#1 link partner capability of 45e1.

Now, perhaps it's just timing out and blocking on various things. I've added an dhcp config for eth1 and an extra putlcd in rcS, and left the machine for 20-odd minutes, but no change. An nmap -sP ping scan over the dhcp range followed by an arp -a to look for MAC addresses didn't reveal the qube, although there is a link light on the hub. In a couple of days I'll pull the disk out and see what got logged.



A couple of minor errors in the documentation - there should be an ext3
filesystem on /dev/hdc6 ready for /home (not strictly necessary to get
the thing to load, but more easily done at install time than later), so
another 'mkfs.ext3 /dev/hdc6' is needed and after mounting /dev/hdc2
user should 'mkdir /mnt/boot' before mounting /dev/hdc1.

-jim

Martin Michlmayr wrote:
* Jim Cheetham <jim@iNode.co.nz> [2004-10-07 12:07]:

Assuming that you don't have time to tarball for a Qube 2700


I've generated a tarball now.  I don't have a Qube without serial
console to test, but it doesn't show any console output on my Qube2 so
hopefully it will work!  Instructions are below.


Preparing the hard drive
========================

Take the hard drive out of the Cobalt machine and put it into any PC.
First, you have to partition the disk.  You *need* a /boot partition as
hda1 as type ext2 revision 0 and a / (root) partition as hda2 as type
ext3.  The installation will by default try to mount swap from hda5 and
home from hda6 but that can always be changed.  The layout I had was the
following (which is the default "desktop" scheme debian-intaller will use
on Cobalt):

   Device Boot      Start         End      Blocks   Id  System
/dev/hda1   *           1         207       97776   83  Linux
/dev/hda2             208        2842     1245037+  83  Linux
/dev/hda3            2843        8912     2868075    f  W95 Ext'd (LBA)
/dev/hda5            2843        3216      176683+  82  Linux swap
/dev/hda6            3217        8912     2691328+  83  Linux

Use fdisk to partition your disk (let's call the disk hdc1 for now,
assuming that you put it on the 2nd IDE controller in your PC).  Make sure
that at least hdc1 is labled as a Linux partition (type 83).  Now format
the drive:

% mkfs.ext2 -r 0 /dev/hdc1
% mkfs.ext3 /dev/hdc2

Mount the disk somewhere:

% mount /dev/hdc2 /mnt
% mount /dev/hdc1 /mnt/boot

Create /home and /proc which are missing

% mkdir /mnt/proc /mnt/home

Downloading and installing the Debian base system
=================================================

Download the compressed base system (about 57M).  This contains a basic
system with a basic set of package of sarge as of 2004-10-13.

% wget http://people.debian.org/~tbm/cobalt/base.tar.bz2
% wget http://people.debian.org/~tbm/cobalt/base.tar.bz2.asc

Verify that the GPG signature matches:

% gpg --verify base.tar.bz2.asc base.tar.bz2

In this base, I have made the following modifications:
 - changed /etc/inittab so it won't use ttyS0
 - patched /etc/init.d/rc and /etc/init.d/rcS so it will show status
   information on the LCD panel

Now untar this system onto the drive:

% cd /mnt
% bzcat ~/base.tar.bz2 | tar -xvf -

Boot the system
===============

Put the hard drive back into the Cobalt and boot it.  It will automatically
load CoLo from disk which will then boot a kernel and start Debian.  You
should see the services on the LCD panel as they are started.  The system
will try to obtain an IP address via DHCP on eth0.  SSH is installed and
remote root logins are allowed.  The password is "root".  There are no
users yet.

What to do now
==============

You have have a working system.  There are a few things you should do,
though:

 - change the root password
 - add normal user accounts
 - regenerate the SSH key (since the private key is included in the base
   system on my web page)
 - Edit /etc/apt/sources.list and use a Debian mirror close to you and
   then type: apt-get update





Reply to: