Re: estrange messages from mpich2 (tight integrated)

Fixed, Reuti:

Even though the image 'main.x' was in the path and accessible and  
executable from all the nodes I interactively logged in, it didn't  
suffice. However, from your suggestions, I explicitly stated the path  
along the executable and it turned out to fixed the problem:
mpiexec -rsh -nopm -np $NSLOTS -machinefile $TMPDIR/machines /home/ 
bta/MPICH2/F90/main.x         (Works)
 instead of

mpiexec -rsh -nopm -np $NSLOTS -machinefile $TMPDIR/machines   
main.x                                                   (Does not)
Thank you, Reuti

Baudilio



On Sep 5, 2007, at 5:25 PM, Reuti wrote:

> Am 05.09.2007 um 22:53 schrieb Baudilio Tejerina:
>
> > Hi ALL:
> >
> > I have just installed MPICH2 (1.0.5) in a cluster (ROCKS) and I'd  
> > like to have all the parallel jobs under GE control. For that,  
> > I've followed Reuti's "demonless" procedure described in
> > http://gridengine.sunsource.net/howto/mpich2-integration/mpich2- 
> > integration.html
> > The script bellow is the one I used for testing the installation  
> > and next is the contents of the error file.
> > The weird messages that i don't know how to interpret nor where  
> > they may come from are those with permission denied.
> > Does anyone have any explanation for this?
> Assuming the path to main.x is in your default $PATH, can you read  
> this file if you login to a node? Is the mountpoint not mounted  
> with "noexec" on the node?
> -- Reuti
>
> > Thank you
> > Baudilio
> >
> >
> >
> >
> > #!/bin/sh
> > #$ -S /bin/sh
> > #$ -cwd
> > #$ -j y
> > #$ -pe mpich2 4
> > #$ -o TEST_MPICH2.err
> >
> > #export MPIEXEC_RSH=rsh
> > #export PATH=/home/reuti/local/mpich2_smpd/bin:$PATH
> >
> > export MPIEXEC_RSH=rsh
> > #Make sure we use the rsh wrapper provided by the PE. A copy of it  
> > will be placed on $TMPDIR:
> > export PATH=$TMPDIR:$PATH
> > echo BTA: my RSH is: `which rsh`
> >
> >
> > export PATH=/opt/local/mpich2-1.0.5/bin:$PATH
> > echo BTA: my MPIEXEC is: `which mpiexec`
> > mpiexec -rsh -nopm -np $NSLOTS -machinefile $TMPDIR/machines main.x
> >
> > exit 0
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > THIS IS THE CONTENTS of the error FILE:
> >
> >
> >
> >
> >
> >
> >
> > -catch_rsh /opt/gridengine/default/spool/compute-0-1/active_jobs/ 
> > 565.1/pe_hostfile
> > compute-0-1
> > compute-0-1
> > compute-0-0
> > compute-0-0
> > BTA: my RSH is: /state/partition1/565.1.all.q/rsh
> > BTA: my MPIEXEC is: /opt/local/mpich2-1.0.5/bin/mpiexec
> > env: main.x: Permission deniedenv: main.x: Permission denied
> > env: main.x: Permission denied
> > env: main.x: Permission denied/opt/gridengine/bin/lx26-x86/qrsh - 
> > inherit compute-0-1 env PMI_RANK=0 PMI_SIZE=4 PM\
> > I_KVS=19B130D14D0B84936F8CE02422E66098  
> > PMI_ROOT_HOST=compute-0-1.local PMI_ROOT_PORT=33703  
> > PMI_ROOT_LOCAL=0 PMI_A\
> > PPNUM=0 main.x
> > /opt/gridengine/bin/lx26-x86/qrsh -inherit compute-0-1 env  
> > PMI_RANK=1 PMI_SIZE=4 PMI_KVS=19B130D14D0B84936F8CE024\
> > 22E66098 PMI_ROOT_HOST=compute-0-1.local PMI_ROOT_PORT=33703  
> > PMI_ROOT_LOCAL=0 PMI_APPNUM=0 main.x
> > /opt/gridengine/bin/lx26-x86/qrsh -inherit compute-0-0 env  
> > PMI_RANK=2 PMI_SIZE=4 PMI_KVS=19B130D14D0B84936F8CE024\
> > 22E66098 PMI_ROOT_HOST=compute-0-1.local PMI_ROOT_PORT=33703  
> > PMI_ROOT_LOCAL=0 PMI_APPNUM=0 main.x
> > /opt/gridengine/bin/lx26-x86/qrsh -inherit compute-0-0 env  
> > PMI_RANK=3 PMI_SIZE=4 PMI_KVS=19B130D14D0B84936F8CE024\
> > 22E66098 PMI_ROOT_HOST=compute-0-1.local PMI_ROOT_PORT=33703  
> > PMI_ROOT_LOCAL=0 PMI_APPNUM=0 main.x
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: users-unsubscribe@xxxxxxxxxxxxxxxxxxxxxxxx
> > For additional commands, e-mail: users-help@xxxxxxxxxxxxxxxxxxxxxxxx
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@xxxxxxxxxxxxxxxxxxxxxxxx
> For additional commands, e-mail: users-help@xxxxxxxxxxxxxxxxxxxxxxxx