Cannot run jobs more than on two nodes

finnishfoam · September 20, 2012, 19:38

We are running OpenFoam with Thirdparty openMPI 1.5.3.
We can only sign jobs upto two nodes.
If we sign jobs to 3 or more nodes job never starts and command just hangs.

This is our command :
(/mirror is NFS share visible on all nodes)

/mirror/OpenFOAM/ThirdParty-2.1.x/platforms/linux64Gcc/openmpi-1.5.3/bin/mpirun -np 12 -hostfile machines /mirror/OpenFOAM/OpenFOAM-2.1.x/bin/foamExe c -prefix /mirror/OpenFOAM interFoam -parallel | tee log

Any ideas why this does not work on more than two nodes (4 cores per node)?

magazoni · December 7, 2012, 09:25

Hi,

Did you resolve your problem? Because it's happening the same thing with my cluster.

Regards,
Felipe Magazoni

duongquaphim · December 8, 2012, 12:41

we use command like

#mpirun -machinefile $PBS_NODEFILE -mca plm_rsh_agent rsh -np 16 interFoam -parallel >& output.$PBS_JOBID.txt

and it works well.

September 20, 2012, 19:38	Cannot run jobs more than on two nodes	#1
finnishfoam New Member Antti Korhonen Join Date: Sep 2012 Posts: 1 Rep Power: 0	We are running OpenFoam with Thirdparty openMPI 1.5.3. We can only sign jobs upto two nodes. If we sign jobs to 3 or more nodes job never starts and command just hangs. This is our command : (/mirror is NFS share visible on all nodes) /mirror/OpenFOAM/ThirdParty-2.1.x/platforms/linux64Gcc/openmpi-1.5.3/bin/mpirun -np 12 -hostfile machines /mirror/OpenFOAM/OpenFOAM-2.1.x/bin/foamExe c -prefix /mirror/OpenFOAM interFoam -parallel \| tee log Any ideas why this does not work on more than two nodes (4 cores per node)?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
First Parallel Run - need some help	Gian Maria	OpenFOAM	3	June 17, 2011 12:08
Help-Edit Run in Progress	BalanceChen	CFX	0	March 30, 2011 09:23
Record Run Times in Fluent	Ravi	FLUENT	3	October 7, 2008 22:41
Cycling a transient run	Chetan Mistry	Phoenics	0	March 3, 2004 09:50
Running 2 CFD jobs on one PC	steve podleski	Main CFD Forum	17	February 16, 2000 14:40

December 7, 2012, 09:25		#2
magazoni New Member magazoni Join Date: Jul 2012 Posts: 11 Rep Power: 13	Hi, Did you resolve your problem? Because it's happening the same thing with my cluster. Regards, Felipe Magazoni

December 8, 2012, 12:41		#3
duongquaphim Member Duong A. Hoang Join Date: Apr 2009 Location: Delft, Netherlands Posts: 93 Rep Power: 17	we use command like #mpirun -machinefile $PBS_NODEFILE -mca plm_rsh_agent rsh -np 16 interFoam -parallel >& output.$PBS_JOBID.txt and it works well.