[Techtalk] progress with debian diskless client

[Chris Wilson]

Hi Maria,

On Wed, 15 Apr 2009, Maria McKinley wrote:

> So, it looks like the machine was probably going through all of the run
> levels, but it was a very stripped down OS, so there wasn't really
> anything to start in the higher run levels (was missing nfs-common among
> many others). I used the chroot command on the server to install stuff
> using dpkg and get-selections so that it would look like the server.
> Apparently there is something installed on the server that nfs booting
> really doesn't like, as when I booted up the new system, I again had the
> problem of booting into a read-only root directory. I do not really want
> to start installing packages one at a time to see what breaks it, so if
> anyone has any ideas for likely candidates or another way to go about
> this, I would love to hear about it.

Does it mount the root filesystem read-write by NFS from the server? If 
so, then perhaps it's trying to mount it read-only first and then run a 
filesystem check? You could symlink fsck and e2fsck to /bin/true in the 
guest system (NOT the host :) to see if that helps? Otherwise, we'd need 
to see complete boot logs from the system, including kernel logs, to 
figure out what's going wrong.

Another approach to this might be to use the LTSP support in Ubuntu, which 
builds a system for network booting diskless clients, so that sounds like 
exactly what you want. The default is that the clients mount their root 
filesystem on a ramdisk, so that it gets destroyed when they shut down and 
never gets sent to the server, but you can change it to "nfs" in the 
lts.conf configuration file.

Cheers, Chris.
-- 
Aptivate | http://www.aptivate.org | Phone: +44 1223 760887
The Humanitarian Centre, Fenner's, Gresham Road, Cambridge CB1 2ES

Aptivate is a not-for-profit company registered in England and Wales
with company number 04980791.