Hello, In our cluster we have a partition configured as Shared=EXCLUSIVE and we are using this plugins for doing the selection:
SelectType=select/cons_res SelectTypeParameters=CR_Core_Memory We do that because we want to be sure that one job receive a full node (16 cores). We have users that wants to submit job array of one core per job. I there a way to submit a job array like that: sbatch --array=1-16 and having all the jobs running on the same time on just one node? If I do that with the current parameters we have, 16 nodes are used, each one having a job of one core allocated. Is to use to use Shared=NO or Shared=YES the only way to achieve that? If yes, is there an other way to ensure that for each normal job a full node is allocated? Thanks
