[K12OSN] A New direction for LTSP: Diskless Remote Boot

[R. Scott Belford]

On Wed, Sep 10, 2008 at 5:28 AM, Warren Togami <wtogami at redhat.com> wrote:


>>  Any idea what exactly LTSP5 Edubuntu is doing wrong?  We're using almost
> all the same upstream sources, although more aggressively newer versions on
> a more frequent basis.


I am not so sure that I would reduce it to anything wrong.  It may have felt
that way to some users(1), myself included, but from a developer's
perspective the implementation of MueKow, aka LTSP 5, into Edubuntu 7.10,
was quite a spectacular feat.  One CD and no network and you still had a
functiong thin-client, or terminal server, lab.

There seemed to be a discrepancy between user expectations, minimum required
hardware specs, documentation, project goals, and the feedback loop between
users and developers.(2)  Many have become spoiled by Eric and Jim and you,
Warren, because you seek solutions from the users at their level on their
mailing lists.  As many are learning - this is rare and should be treasured.

If I had my way, I would put a few of you on a plane, bring you to Hawaii,
and work it out.  This is Vagrant's comment about rolling out LTSP 5 with
debian-edu(3).  The two of you may want to compare notes, Warren.  Not
unlike the impressive, Red Hat sponsored K12OSN list, debian-edu has a user
mailing list with an awe-inspiring feedback loop between users and
developers.

>
> We are doing X detection stuff in a completely different (and less broken)
> manner.  But otherwise our stuff really isn't that different.


X detection could help.  Using Edubuntu 7.10 with older clients that worked
with 7.04 and the K12LTSP required lts.conf tweaking.  This was Lesson 1 in
appreciating how many headaches Eric and Jim silently saved us.  Our
migration of Peter's lab last year from the k12ltsp to edubuntu 7.04, and
then to 7.10, was quite a Live lesson in upstream integration.

In general, it seems that accommodating older hardware is the challenge with
LTSP 5.  I realize the impressive benefits of sharing one set of maintained
applications, ssh rather than nfs, and what amounts to a virtualized OS that
can be shared amongst many clients.  Seems like Sun is taking Virtualbox
there.

The temptation is to rationalize the newer and faster hardware required
because, as Robert observed(4), we are close to using solid state network
appliances.  The power savings alone may justify this path.  This is the
rub.

As I will witness next month when Apple pays to haul many containers of
hardware away from Hawaii, there is always someone, I think, who can use a
donated computer or a functioning computer lab with donated, diskless
workstations.(5)  All of us on this list know this.  Perhaps the path is for
LTSP 5 to evolve with faster hardware in mind while projects like DRBL, that
let me Etherboot Puppy Linux, address those of us using 'legacy' hardware.

--scott

(1)
https://lists.ubuntu.com/archives/edubuntu-users/2007-December/002964.html

(2)
https://lists.ubuntu.com/archives/edubuntu-users/2008-September/004435.html

(3) http://lists.debian.org/debian-edu/2008/09/msg00065.html

(4) https://www.redhat.com/archives/k12osn/2008-September/msg00156.html

(5) http://www.youtube.com/watch?v=QjBCKRcHAA4
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/k12osn/attachments/20080910/1b33abcc/attachment.htm>