[K12OSN] Seeking advice and help combining K12LTSP, OpenMosix and ThinStation to create one large project

[Jon Spriggs]

Hi all,

Over the past year or so, I've been very interested in the Thinstation
and K12LTSP projects, and in the last couple of weeks, the OpenMosix
project has interested me even further...

I know that there were HowTo: documents for incorporating OpenMosix
into K12LTSP version 3.x, but I know that K12LTSP is now on version
4.2 - will this make a big difference?

I've noticed that OpenMosix works with the 2.4.26 kernel, but K12LTSP
is based on Fedora Core 3 - which uses a 2.6 based kernel. I'm not a
kernel hacker (infact, I've only ever had one successful
kernel-build), so I don't really know how much difference there is
between a 2.4 kernel and a 2.6 kernel. Would there be much to gain (or
lose?) from regressing the K12LTSP 2.6 based kernel to a 2.4 kernel?

I know that for OpenMosix to work correctly, it needs the same kernel
version between all kernels on the cluster - therefore, I presume I'd
need to build the same kernel for the Thinstation project (so chosen
because I may be using both Windows and Linux terminal servers on my
estate). Is it particularly difficult to build a kernel with OpenMosix
in for Thinstation? Would it be worth approaching the Thinstation
developer team about building an OpenMosix based package? Maybe even
asking them to include the OpenMosix packages into the Kernels they
distribute.

Essentially, what I'm not looking to do here is build a new project
from scratch - I'm far too disorganised and lose my focus a little too
quickly for that, but what I'd like to do is organise three mutually
compatible projects into one core product which can be used to help
the educational world teach and learn, and to help the business world
work smarter with the increased drive towards server based computing.

I know this work is already done as individual projects, and K12LTSP
already does some of the work that Thinstation does, but Thinstation
allows you to have a local filesystem, without being overly dependent
on DHCP and TFTP servers. Maybe I'm being a little overly ambitious,
but I think this could be a really great project.

I'd welcome any comments, positive or negative - especially from
anyone who may have done any work along the same lines, or who can
suggest alternative projects which may also fall well into the same
tree?

-- 
Jon "The Nice Guy" Spriggs