Hi, You could try something like: srun -w "mybigmemorynode","mybigmemorynode2","mynodewithlessmemory"
Le 19/10/2015 21:44, Ghislain LE MEUR a écrit : > Problem to run a job with more memory only on the node where the job > start > Hello, > > I have a pool of nodes with the same memory size (128G) and 2 others > with more memory size (512G) for other kind of softwares. > > Is it possible to start a job on one of these 2 nodes with 512G and > also compute on others nodes with 128G of memory ? The job needs only > more memory on the first node where the job start. > > if I run "srun/sbatch --mem 512" the job will failed because there is > not enough memory on others nodes. > > I have played with the Prolog/PrologSlurmctld variable to try to > define RSS limits without success because prolog script is not the > parent of surmstepd. > > Anyone have an idea ? > > Regards, -- --- Mehdi Denou Bull/Atos international HPC support
