Fatal error when running on HPC through slrum

GROMACS version:2018.1
GROMACS modification: NO
Here post your question

Hello everyone,

I’m trying to optimize a cluster for the best performance. I have used a .tpr file, which was simulated for 1 microsecond in another system. Now, when I run the .tpr in the current HPC, I’m getting an error as follows:

Fatal error:
88 particles communicated to PME rank 4 are more than 2/3 times the cut-off out of the domain decomposition cell of their charge group in dimension x. This usually means that your system is not well equilibrated.

I have changed node 4 to 1 and altered other parameters in the SLURM script:

#SBATCH --nodes=1
#SBATCH --ntasks=6
#SBATCH --tasks-per-node=6

Now, I have created the simulation system from the HPC with the same structure and minimized to a maximum force of <1000, and the simulation is equilibrated—all these on a single node. However, when I run the simulation without the SLURM script, on the single node, it runs without any issue. Nonetheless, even for a single node, the issue persists when I run it through SLURM.

Can someone have a look at this and provide any suggestions?

The relevant part of the SLURM script and log file is here:

:

Hi,

maybe you are running into a bug that was fixed in v. 2022.2 (but I am unsure about whether this bug was already there in 2018.1).

The issue is described here:

You may want to upgrade to 2022.2 or later and see whether that solves your problem!

Best regards,
Carsten