Installing Scyld: some questions

[Daniel Ridge]

On Wed, 11 Apr 2001, Bruno Barberi Gnecco wrote:

> > Running X on the slaves doesn't make a whole lot of sense to me.  With
> > the beowulf setup, you want to just log into the head node and not have
> > to worry about logging into the slave nodes.  This is even more
> > prevelent with the Scyld setup, where the slave nodes don't have any
> > binaries on them, instead you run processes on them by using BProc to
> > propagate the process from the head node to the slave nodes.
> 
> 	We need to run X because we are interested in graphic output.
> We'll be primarily using OpenGL, which (AFAIK) can only be used under X,
> but other libraries (Performer, OpenInventor, GGI, even things like GTK)
> will also be used. X also is sure to provide drivers for our video boards.
> I don't need (and don't want) people logging in the slaves. My idea was to
> run X by a dummy user.
> 
> 	Would you advise against Scyld in this case? I like it, and it
> fits very well for our purposes. Would it be too difficult to install
> X in the hard disk and run in remotely?

I sometimes run X servers on Scyld nodes.

It's basically straightforward and looks like

bpsh 0 X

The magic detail is that I usually have to copy the Xconfig file over
there and that I run a separate copy of the X font server on the frontend
that listens on a TCP port.

Other than that -- no problems.

I use this at home for the Scyld node that drives my television and plays
my MP3s. I have always envisioned that running Scyld 'backwards' so that
each node actually uses BProc to host a small-footprint interactive
environment. Sort of a cross between the Sun SunRay and LTSP projects.

Although program binaries don't live on the Scyld nodes, you are welcome
to use the local disks for any kind of data file or even filesystem you
wish. The fact that Scyld doesn't require the disks at all on the nodes
gives you supreme flexibility.

Regards,
	Dan Ridge
	Scyld Computing Corporation