We just bought single node server with 2x18=36 cores and 200 Gb ram for scientific calculations. There are several users and it is very hard to track jobs and to schedule them between users. This is my first time i am administrating server.
From my experience using supercomputers there are workload schedulers like PBS and SLURM.
Is it possible to configure SLURM for single node mashine with some kind of virtual login node lets say with 4 cores and assign multiple jobs with different number of processors, like running two jobs each with 16 cores.
If so please give me hints how to do that :).