Scyld Beowulf

Erik Paulson epaulson at students.wisc.edu
Sun Dec 3 15:12:54 PST 2000


On Thu, 30 Nov 2000, Dan Smith wrote:
> Date: Thu, 30 Nov 2000 14:51:08 CST
> To: beowulf at beowulf.org
> From: Dan Smith <dcs at iastate.edu>
> Sender: beowulf-admin at beowulf.org
> Subject: Scyld Beowulf
> 
> I was recently put in charge of setting up a linux cluster that we
> purchased
> as a package from Atipa.  Each node is running RH 6.2 except the master
> node
> which is running RH 7.  Currently, there is no MPI, no Beowulf, no
> queuing
> system installed.  The group that owns this cluster wants to run the same
> Monte Carlo sims on each node but with different input files.  No MPI or
> the
> like is required for that.  These are serial codes.  We would like to be
> able to log in to the master node only, and send the jobs out to the
> slave
> nodes, i.e. via some queuing system, but we will eventually want to run
> MPI
> parallel codes as well.  Is the Scyld Beowulf package all I need to do
> both,
> or will I need a separate queuing system, i.e. DQS, to control spreading
> the
> serial Monte Carlo codes to the slaves?  Is there another alternative?
> 

<Shameless plug>
Condor, from the University of Wisconsin. http://www.cs.wisc.edu/condor/ 
</Shameless plug>

It's not clear to me from the Scyld site what queueing system they include,
if any. But the Scyld folks are
pretty good and if they did include one it'll handle everything you're
asking for.

-Erik

> ---
> Daniel C. Smith                  |    Iowa State University
> Graduate Assistant               |    Department of Physics and Astronomy
> dcs at iastate.edu                  |    Ames, IA 50011
> 
> _______________________________________________
> Beowulf mailing list
> Beowulf at beowulf.org
> http://www.beowulf.org/mailman/listinfo/beowulf




More information about the Beowulf mailing list