Huge increase of RAM makes parallel mpirun fails

choch · November 8, 2017, 10:34

Hello,

I am trying to launch a 2D simulation in my University cluster which runs on SLURM workload manager. My case is quite simple, the mesh being regular and homogeneous.

I tested it on my local machine and there is no problem but when running in the cluster, the RAM hugely increases leading to the fail of the work "slurmstepd: error: Exceeded job memory limit at some point."

The available memory is 45 GB. In my local machine only 2 GB are used anyway. Now the weird thing is that I was so desperate so I tried to run a 2,500,000 cells case which was successful. This randomness is a bit disorientating. Could it come from the MPI settings of the cluster ?

Any one had the same issue ?

I thank you all for your time and hope you can provide me with some help,

Charlie

be_inspired · November 11, 2018, 17:38

Here the same problem. Did you find how to solve it?

EJohn · November 12, 2018, 11:42

If your case file runs perfectly on your local machine, then there must be an issue with either your SLURM script file or the MPI settings on the cluster.
I would first double check my script files to make sure my code is correct and then try asking your University cluster administrator for help.

November 8, 2017, 10:34	Huge increase of RAM makes parallel mpirun fails	#1
choch New Member Charlie Barraud Join Date: Jun 2016 Posts: 5 Rep Power: 10	Hello, I am trying to launch a 2D simulation in my University cluster which runs on SLURM workload manager. My case is quite simple, the mesh being regular and homogeneous. I tested it on my local machine and there is no problem but when running in the cluster, the RAM hugely increases leading to the fail of the work "slurmstepd: error: Exceeded job memory limit at some point." The available memory is 45 GB. In my local machine only 2 GB are used anyway. Now the weird thing is that I was so desperate so I tried to run a 2,500,000 cells case which was successful. This randomness is a bit disorientating. Could it come from the MPI settings of the cluster ? Any one had the same issue ? I thank you all for your time and hope you can provide me with some help, Charlie

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[snappyHexMesh] Error while running SnappyHex in parallel	mg.mithun	OpenFOAM Meshing & Mesh Conversion	1	February 10, 2016 14:13
MPIRUN fails	lfbarcelo	OpenFOAM	3	March 29, 2010 08:41
1.6.x -> snappyHexMesh in parallel fails ...	podallaire	OpenFOAM Bugs	11	August 4, 2009 14:57
Does it metter to increase RAM for solver speed??	raima	Main CFD Forum	1	February 28, 2008 11:47
Serial run OK parallel one fails	r2d2	OpenFOAM Running, Solving & CFD	2	November 16, 2005 13:44

November 11, 2018, 17:38		#2
be_inspired Senior Member M. Montero Join Date: Mar 2009 Location: Madrid Posts: 155 Rep Power: 17	Here the same problem. Did you find how to solve it?

November 12, 2018, 11:42		#3
EJohn New Member Edward Join Date: Nov 2018 Location: Ottawa Posts: 9 Rep Power: 8	If your case file runs perfectly on your local machine, then there must be an issue with either your SLURM script file or the MPI settings on the cluster. I would first double check my script files to make sure my code is correct and then try asking your University cluster administrator for help.