You might want to have a gander through this short "getting started
with LAM/MPI" tutorial:
http://www.lam-mpi.org/tutorials/lam/
On Jul 27, 2007, at 10:59 AM, Deanna DiBernardi wrote:
> I am trying to boot a cluster through LAM, so to do so, edited the
> lam-bhost.def
> file. I manage to lamboot all nodes in the cluster, but when I use
> a command
> such as:
> mpirun n0 n1-4 <filename>
> -or-
> lamexec n0 n1-4 <filename>
> the program seems to boot all on the first node, ie, runs the same
> program 5
> times in a row on the main screen. this occurs, or once i received
> an error
> message saying that the IP address 127.0.0.1 (localhost) (naming it
> n0) would
> not accept. However, I have set my private network IP addresses in the
> 192.168.xx.yy range. Why is it reading the localhost as n0 if I've
> lamboot-ed
> the cluster from my lam-bhost.def file, where all the other nodes
> are named?
> And how do I fix it?
> Thanks so much!
> Deanna
> _______________________________________________
> This list is archived at http://www.lam-mpi.org/MailArchives/lam/
--
Jeff Squyres
Cisco Systems
|