Jess,
Thanks my script is also quite similar. I still think lamboot isn't
successful, you should get the message "topology done" in the o/p right? I
don't get it.
This is my pbs script:
#!/bin/sh
#PBS -N matlab-test
#PBS -l nodes=2:ppn=2
#PBS -l Walltime=00:05:00
#PBS -A pushkar
#PBS -m b
#PBS -m e
#PBS -m a
#PBS -mea
#PBS -r n
#PBS -V
# Tell the PBS system to cd to this directory
cd $PBS_O_WORKDIR
# cp $PBS_NODEFILE mpidefconf.m
lamboot -v -d $PBS_NODEFILE
# Now you are done with PBS commands. Start your MPI program.
# This program required 10 processors.
# matlab -nosplash -nojvm < Main.m
lamhalt
Thanks,
Pushkar
> -----Original Message-----
> From: lam-bounces_at_[hidden] [mailto:lam-bounces_at_[hidden]]On Behalf
> Of jess michelsen
> Sent: Tuesday, June 01, 2004 4:52 PM
> To: General LAM/MPI mailing list
> Subject: Re: LAM: unable to boot
>
>
> Hello Puskar!
>
> Heres a sample submit script, that I use under PBSpro (if you don't have
> the task manager, just omit -ssi boot tm. Under PBS, the node-list is
> parsed to the lamboot command in the line NPROCS=`wc -l < $PBS_NODEFILE`
> Apart from the preamble not seen in your snippet (is it missing?), I
> assume you should have an mpirun line in which matlab would appear as
> the executable. I'm also surprised that matlab exists in an mpi-parallel
> version.
>
> #!/bin/sh
> #PBS -N jobname
> #PBS -r n
> #PBS -q queuename
> #PBS -l nodes=nodecount
> #PBS -l walltime=hours:00:00
> #PBS -j oe
> NPROCS=`wc -l < $PBS_NODEFILE`
> lamboot -b -ssi boot tm
> mpirun -O -ssi rpi tcp -np $NPROCS -wd $PBS_O_WORKDIR execname [options]
> lamhalt
>
> Best regards, Jess Michelsen
>
> On Tue, 2004-06-01 at 22:48, Tim Mattox wrote:
> > Hello Puskar,
> > With the LAM 7.0.x series, the lamboot command requires each commandline
> > option to be listed separately. An annoyance, but nothing too bad.
> > Change your "lamboot -vd" to "lamboot -v -d" in your PBS script.
> >
> > You also seem to be missing the filename with a list of hosts for
> > lamboot,
> > but I don't use PBS, and am not sure what to use to get PBS's list of
> > allocated machines.
> >
> > Hmmm, maybe I should let others answer your question, since it seems you
> > have some odd mixture of things... matlab, to my knowledge, is not an
> > MPI
> > program...
> >
> > On Jun 1, 2004, at 4:25 PM, Pushkar Pradhan wrote:
> > > I keep on getting these errors when trying to run lamboot
> (I've pasted
> > > the
> > > errors at the end).
> > > recon has completed successfully.
> > > This is how I boot them (from my pbs script):
> > > # Tell the PBS system to cd to this directory
> > > cd $PBS_O_WORKDIR
> > > ps -aux
> > > lamboot -vd
> > > # Now you are done with PBS commands. Start your MPI program.
> > > matlab -nosplash -nojvm < cluster_test.m
> > > lamhalt
> > >
> > > The ps -aux command doesn't show lamd, how do I make sure lamd starts
> > > or did
> > > I miss it somehow?
> > >
> > > ERRORS:
> > > Redstone[63] pushkar$ m matlab-test.e67443
> > > Synopsis: lamboot [-b] [-d] [-h] [-H] [-l] [-s] [-v] [-V]
> [-x] [-nn]
> > > [-np]
> > > [-c <conf file>] [-prefix </lam/install/path/>]
> > --
> > Tim Mattox - http://homepage.mac.com/tmattox/
> > tmattox_at_[hidden] - http://aggregate.org/KAOS/
> >
> > _______________________________________________
> > This list is archived at http://www.lam-mpi.org/MailArchives/lam/
>
>
> _______________________________________________
> This list is archived at http://www.lam-mpi.org/MailArchives/lam/
|
