Re: [SLUG] Linux Kiosk @ USF

From: Ian C. Blenke (icblenke@nks.net)
Date: Fri Feb 10 2006 - 16:27:35 EST


Richard Morgan wrote:

> It sounds like you're talking about using consoles to connect via
> serial port. That's certainly possible and used to be done a lot 'back
> in the day'. To
> connect that many terminals, you'd need a multiport serial card to
> expand the
> machine from the usual one or two serial ports.
>
> Quoting Tevfik Yücek <tevfik.yucek@gmail.com>:
>
>> I am a student in engineering. Our unix system admin set up 3 kiosk
>> computers running linux (red hat?). They are for a similar purpose as
>> your application.
>>
>> While chatting with him, he mentioned that only one computer can be
>> connected to 3 monitors and do the same thing. He said he just did not
>> have time to exploit it. I have no idea how you can do it but it would
>> be much easier and better if you just use 1 computer and 10 displays.
>> Maybe some other people from the list has a clue on this. I think it
>> is worth to try.
>

Yes, you can hook up more than one monitor to a PC.

Moreover, with USB keyboards and USB mice you can effectively run
separate "consoles" fairly painlessly.

This is called "multi-seat" operation for a "multi-head" configuration,
and is easy to search for via Google.

 From my experience, if you were to attempt this yourself, I highly
suggest running _one_ Xserver to service all of the multi-headed
consoles. Running more than one Xserver at a time requires some fun 2.6
kernel hacking to make it VT safe for multi-head. This is all part of
the "backstreet-ruby" project (nothing to do with the Ruby language,
btw). Boy is it a royal PITA.

 From what Tom is asking about at the moment though, I think that the
LTSP approach is a pretty good one.

If it were me, I would:
1. Build the LTSP server to securely handle all of the user desktops in
the same way.
2. Build a auto-configuration kernel (NFSROOT, etc) that is appropriate
for all of the available hardware.
 - one kernel might not be adequate, you may have to tag specific
netboot clients to receive specific kernels for that PC's hardware.
3. Setup machines to PXE boot on the segment from a boot server.
 - if a machine can't PXE boot, I would use a grub bootfloppy with an
etherboot appropriate to PXE bootstrap for that machine.
 - if a machine doesn't have a boot floppy, burn an el-torito ISO image
to do the same.
 - if a machine doesn't have that, install a master boot record on the
harddrive to grub boot to the same

This is, of course, assuming that you control DHCP on those segments, or
can otherwise segment the boxes in a way where you can respond to the
PXE DHCP lease requests.

If PXE isn't possible, I'd probably start devising boot media for each
machine to statically config that machine to network boot off of the
LTSP server.

Once you have the PCs booting and pulling up sessions from the LTSP
server, most of the physical labor is done. At this point you would be
fiddling with getting things like printing to work and otherwise
configuring user sessions for each station.

 - Ian C. Blenke <ian@blenke.com> http://ian.blenke.com/

-----------------------------------------------------------------------
This list is provided as an unmoderated internet service by Networked
Knowledge Systems (NKS). Views and opinions expressed in messages
posted are those of the author and do not necessarily reflect the
official policy or position of NKS or any of its employees.



This archive was generated by hypermail 2.1.3 : Fri Aug 01 2014 - 18:17:30 EDT