[Discuss] FreeNx

[Alan W. Irwin]

On 2007-11-22 08:23-0800 pw wrote:

> John Blomfield wrote:
>
>> Thanks, I am beginning to get my mind around all of this.  I have set 
>> about trying out all the options but I have been unable to get freeNx and 
>> 2X servers to work.  Probably a combination of bugs in the distro 
>> implementations and my incompetence. However, I have got plain old vnc 
>> working perfectly. I found it relatively simple compared to the others and 
>> a wealth of "how to's" around to help.  As yet I have not been brave 
>> enough to try your X server solution.  I realise that vnc is not secure 
>> but that hardly matters in my limited home environment and I have an 
>> external firewall router.
>> 
>> John Blomfield
>
>
> If you don't wish to mess with the setting up the vnc server resolution
> for each client you'd best get Alan to run you through the whole X thin
> client thing. He does that stuff in his sleep.

Probably... :-)

Actually anyone can grok remote use of X clients in their sleep if they
remember just one principle; X has a classical client/server architecture
where the X server is always run on the local computer that has the
monitor/mouse/keyboard attached while the X clients (e.g., KDE or GNOME
applications) can be run on that computer or any other remote computer that
is accessible over the net.

In the typical case of a Linux PC or laptop the X clients run on the same
machine as the X server so people mentally merge the clients with the server
and forget that they are really completely separate.  More confusion is
caused by people's preconception that a "server" is some honking machine as
opposed to a particular kind of software (and in the X case, the X server
software can and often does run on the smallest machines available and
not necessarily on the "server" machine).

So, John, nail that above principle to your wall, and remote X clients
become really easy to both understand and use.

Let's say you have a box called chickadee (a slow 133MHz/64MB box would do
fine) where the X server runs and a high-speed box called raven where the X
clients run.

Then on raven, configure the xdm X client to only respond to queries from
chickadee.  (This is the "tough" part which takes maybe a few minutes.) On
chickadee configure X like you normally do (or leave it alone if it is
already configured as in the normal case).  Then simply launch the X server
on chickadee by running

X -query raven

The chickadee user then gets a login prompt from xdm running on raven, and
from then on they are using the raven desktop (KDE in our case) at the same
speed as if they were sitting at the raven keyboard.

Running remote desktops by taking advantage of the separation between remote
X clients from the local X server works so well that people think there must
be some complicated trick behind it.  However, it is just a nice byproduct
of a great choice of software architecture for X as summarized in the above
principle, and to set it up just takes a few minutes with configuring xdm on
the box where the X clients (the desktop software such as GNOME or KDE apps)
will be running.

Alan
__________________________
Alan W. Irwin

Astronomical research affiliation with Department of Physics and Astronomy,
University of Victoria (astrowww.phys.uvic.ca).

Programming affiliations with the FreeEOS equation-of-state implementation
for stellar interiors (freeeos.sf.net); PLplot scientific plotting software
package (plplot.org); the libLASi project (unifont.org/lasi); the Loads of
Linux Links project (loll.sf.net); and the Linux Brochure Project
(lbproject.sf.net).
__________________________

Linux-powered Science
__________________________