Clusters run with the Slurm workload manager (submitting jobs with mpirun is not allowed unlike on single instances). If you are unfamiliar with this scheduler, there are many online tutorials and blogs in addition to the own Slurm website (https://slurm.schedmd.com/documentation.html). To simplify matters, there are several Slurm scripts in the BATCH subdirectory ready for job submission, which can be submitted with $ sbatch WRF_256.sbatch or $ sbatch WRFGrav_256.sbatch (for Graviton2 clusters). This command will run the case located at /home/centos/WRF-4.2.2/test/em_real with 256 MPI ranks.
Deployment of AWS cluster with custom odyhpc AMIs
(iv – Submitting jobs to the cluster scheduler)