SLURM v1.2 is four years old and I would not recommend its use. I believe SLURM's ability for nodes to be in more than one partition was added later.
Quoting "Löhnhardt, Benjamin" <benjamin.loehnha...@med.uni-goettingen.de>: > Hi, > > we are using Slurm (1.2.25) and LSF on our cluster system. Is it possible to > share resources over more than one partition? > > We already have a lsf partition for all lsf jobs. Now we are > thinking about to > create another partition for MPI jobs, which are directly send to Slurm via > srun. In our first tests it was possible to start a MPI job via the mpi > partition although some of the nodes already are in use. The other way around > is not possible, so LSF jobs will not be executed while a MPI job > occupies the > resource. > > Is there an error in our configuration or isn't it possible to share > resources > over partitions? > > Regards, > Benjamin >
