[Galactus]

Greetings Everyone,
I'm using a cluster to run my OpenFoam simulation. I am using

mpirun -np 24 pimpleDyMFoam -parallel >& log

Then

tail -f log

to keep track of the simulation. But the simulation stops at random places like after the "smooth solver" line and takes lot of time to proceed further only to stop again.

My time steps are in the order of "e-7" and it takes around 5 minutes to reach 1e-4 from 0. Hence it is very fast but stops randomly and takes hours before it continues on its own.

I've previously run a similar case in my PC and did not face any such issue. The only difference between the cluster and PC is that the mesh is much more refined on the cluster.

Any suggestion or tips on how to encounter this issue?

[Galactus]

Greetings Everyone,

I let the solver run for 24 hours on the cluster and it stopped at two different places/timesteps (in-between lines smooth solver & GAMG). The report which is generated by the cluster shows a total run time of 24 hours but only 55 minutes of CPU usage time. There was no shortage memory (Hard Disk/Ram) either.

I took the same case and ran it on my PC but with 8 processors and it did not have any issue (though it is slow and is still running). Is there anything extra required to run the program on cluster? The cluster has modules built-in and we can load them to run our programs. Any insight would be appreciated!