PXE netboot install attempt report (2003-05-06)
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Hello everyone!
I'm back with another report. This time (and probably from now on) I
installed via PXE using the netboot image from 2003-05-06. Unfortunately,
this installation did not complete -- ie. I was not able to hack my way to
an installed system. Here are the issues I encountered:
1) Configuring the network using DHCP failed. The image had netcfg-dhcp
0.22 installed, and it reported very simply:
No dhcp client found. I cannot continue.
This package requires pump or dhcp-client.
I worked around this by configuring using a static IP. It seems kinda bad
if there is no DHCP client on the image -- especially for a netboot
oriented image. Anyone know why this is?
2) Choosing a keyboard layout failed. For no particular reason that I
could discern (no messages on the console, tty3, or tty4) kbd-chooser
[0.20] decided to segfault. Here's the error message:
Segmentation fault
kbd-chooser's postinst exited with status 35584
I just ended up skipping this step entirely and proceeded to
hw-detect-full.
3) Couldn't create a swap partition. As reported ad nauseum, mkswap is
once again missing from the image. Can anyone say why the image is back
to using busybox instead of busybox-cvs? Is there something wrong with
- -cvs to cause us not to use it?
4) Another oldie but goodie: base-installer [0.020] can't find
/target/etc/fstab (see bug #186443). I'm now of the opinion that this is
base-installer's fault -- why does it assume/require that the file exist?
The way I (now) understand things, the fstab is created in the
prebaseconfig step (by the script /usr/lib/prebaseconfig.d/40fstab) which
doesn't run until right before the reboot. I think that base-installer
should check for /target/etc/fstab before running debootstrap, but not die
if it's not there yet. I will be filing a bug (w/patch) unless I hear
something convincing to the contrary.
5) Last, but certainly not least, the show stopper -- kernel installer
[0.020] fails to install the kernel. It exits with a status 5120 after
printing the following:
kernel-installer: info: Setting do_initrd='yes'.
kernel-installer: info: Installing kernel 'kernel-image-2.4.20-1-386'.
kernel-installer's postinst exited with status 5120
Digging around, I found the following error messages on tty3 (and in
/var/log/messages):
/usr/sbin/mkinitrd: Cannot determine root device
Failed to create initrd image.
dpkg: error processing kernel-image-2.4.20-1-386 (--configure):
subprocess post-installation script returned error exit status 25
Errors were encoutered while processing:
kernel-image-2.4.20-1-386
E: Sub-process /usr/bin/dpkg returned an error code (1)
Anyone know what might cause such breakage in mkinitrd? Is it a problem
with the kernel-image or with mkinitrd itself?
All in all, it's not so bad. Two problems (1,3) are related to the image
itself, #4 is a simple patch to base-installer.postinst, and the other two
are real (new?) problems. Let me know if you want me to file bugs for the
last two, and against which packages they should go (2 is easy, 5 I don't
know).
Well, that's all for tonight folks, thanks for reading! Stop by next week
for another edition of "Joe tests the debian-installer": Same bat-time,
same bat-channel ;-)
Joe Nahmias, DD wannabe
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (GNU/Linux)
iD8DBQE+uIiOKl23+OYWEqURAjFMAJ4nm3gDJaRhFE/h4faHilfsv7kdtgCgxGyv
ykX9zKCkhfmQnjU1h122B1o=
=8eMy
-----END PGP SIGNATURE-----
Reply to: