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

Bug#256572: Debian Installation Report

Package: installation-reports


Debian-installer-version: 20040616 ftp.debian.org
uname -a: Linux debian 2.4.26-1-686 #1 Sat May 1 18:04:05 EST 2004 i686 GNU/Linux
Date: 20040625
Method: How did you install?    debian-installer with CDROM, #1 & #2
 What did you boot off?        CD
 If network install, from where? no

Machine: 2 machines: PPro[200MHz] 3 disks + CD, and PIII[800MHz] 2 disks + CD
Processor: PPro & PIII
Memory: 96M & 256M
Root Device: ide: hdd & hda
Root Size/partition table: 1G, /dev/md0

Base System Installation Checklist:
[O] = OK, [E] = Error (please elaborate below), [ ] = didn't try it

Initial boot worked:    [O]
Configure network HW:   [O]
Config network:         [E]
Detect CD:              [O]
Load installer modules: [O]
Detect hard drives:     [O]
Partition hard drives:  [O]
Create file systems:    [O]
Mount partitions:       [O]
Install base system:    [E]
Install boot loader:    [E]
Reboot:                 [O]


The intent was to get a machine with all filesystems on RAID1, including
root, swap and boot.

After installing from CD, the network is not configured, and the file
/etc/networking/interfaces must be changed by hand to list the interfaces there.
Although the network is not used in the installation itself, it should be
configured correctly anyway. ["dpkg-reconfigure netabse" dosn't do anything!]

I have seen the same kind of troubles described in Bug#251905.
My workaround was somewhat different, and I did not use LVM.
When the screen "Unable to install selected kernel" is on vt1, I have
installed 'mdadm' in the target chroot system, did some cleanup
[rm -fr /usr/bin/awk /lib/modules] to allow debootstrap to work again
and installed the base-system in a dirty filesystem for a second time.

Eventually, I have skipped installing grub because grub dosn't know about /dev/md*

The raid configuration should also allow to use 'missing' devices,
because the number of drives cannot be changed later and I do not
want to destroy the filesystem that was working OK without raid,
before configuring and testing with the new raid installation is OK.
Only then I want to hotadd the old disk and have the benefits of RAID1.

There are only a few 'small' steps needed and the warning about debian
not supporting RAID on root and boot can be removed:
- If RAID is configured, 'mdadm' must be installed before the kernel
 is configured.
- grub needs to learn about raid partitions
- debootstrap must allow redoing its work on a dirty filesystemi
       [scripts/sarge:167  ln -fs mawk $TARGET/usr/bin/awk]

The Raid partitioning dialogue allows you to define a new partition table
on a raid device.  I have doubts about this being supported by the kernel.

This may need attention of someone:
Jun 26 23:57:10 (none) syslog.warn klogd: md: parted_server(pid 17876) used obsolete MD ioctl, upgrade your software to use new ictls. Jun 26 23:57:20 (none) syslog.warn klogd: md: find-partitions(pid 20656) used obsolete MD ioctl, upgrade your software to use new ictls.

Line 25 in file /etc/console-tools/config should be a comment.
Othewise a kernel parameter "vga=ext" will be reset....

----------------------------------------------------------------  -o)
Matthijs Melchior                                       Maarssen  /\\
mmelchior@xs4all.nl                                  Netherlands _\_v
---------------------------------------------------------------- ----

Reply to: