> I would like to ask the group to discuss the
> possibility that what OP wants is overkill for a local elementary
> school, as well.
To clarify, here are the requirements for this project:
1. 2 LTSP servers per classroom, a primary and secondary.
2. The two servers split the boot and application load from the
terminals. (a nice option: allows us "cluster" slower servers)
3. In the event one server fails, the client sessions on that server
are lost, but the other server assumes the full load. (this is
essential: we're using old, retired gear)
4. No client data will be stored on the LTSP servers... so no server
sync'ing will be required. The logins will authenticate via the
school's existing NT network (via pam_smb or winbind) and users will
be mapped to their NT home directory with Samba.
5. Integration of openMosix with LTSP for compute intensive
classroom applications. (this is complete and working great!)
http://users.adelphia.net/~netdxr/projects/linux/20020912/
6. The finished architecture must be easy to duplicate and
maintainable by educated but "average" people. A failed server must
be removable by a non-technical person without disrupting the still
working system. (remote schools probably won't have Linux guru on
hand or a LUG nearby!)
While the reliability of this system doesn't have to be perfect, it has
to be *much* better then the single point of failure that an old
recycled server represents. So if we can't achieve a reasonable degree
of high availability with this architecture, I won't recommend that we
deploy it to other schools. The catastrophic failure of a classroom full
of terminals wouldn't be a big deal if this wasn't an Linux project...
but there are a lot of eyes on the Penguin right now.
A number of us have invested a lot of our personal time in this
project... but there's more at stake then that. We've also made
significant progress in starting to change the Microsoft mindset and
status quo in our school system. We need to maintain this initiative
and momentum!
I very much appreciate all the support and suggestions I've received
from my original posting... thanks to *everyone* who responded! Per some
excellent suggestions from Hans Ekbrand, I've done some research on the
failover and load balancing features of DHCPd. The most recent version
(3.0.1 RC 9) of ISC's DHCP is supposed to support failover and load
balancing. Some of the postings I've seen aren't very encouraging... but
I'm going to give it a try.
Last night I set up two identical LTSP servers with 3.0.1 RC 9 and am
going to configure them for failover and load balancing. Here are some
useful notes I found if anyone else is interested in giving this a try:
http://www.andrew.cmu.edu/~kevinm/dhcp/failover.html
If I get this working, I'll post my results.
Best regards,
-Tom
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_____________________________________________________________________
Ltsp-discuss mailing list. To un-subscribe, or change prefs, goto:
https://lists.sourceforge.net/lists/listinfo/ltsp-discuss
For additional LTSP help, try #ltsp channel on irc.openprojects.net
