展开

Why are my slurm job steps not launching in parallel?

最后发布时间 : 2022-08-04 17:05:02 浏览量 :

https://stackoverflow.com/questions/68191801/why-are-my-slurm-job-steps-not-launching-in-parallel

#!/bin/bash

#SBATCH --ntasks=2

srun --ntasks=1 sleep 10 & 
srun --ntasks=1 sleep 12 &
wait
srun: Job 247 step creation temporarily disabled, retrying (Requested nodes are busy)
srun: Step created for job 247

Depending on the Slurm version you might have to add the --exclusive parameter to srun (which has different semantics than for sbatch):

#!/bin/bash

#SBATCH --ntasks=2

srun --ntasks=1 --exclusive -c 1 sleep 10 & 
srun --ntasks=1 --exclusive -c 1 sleep 12 &
wait

https://stackoverflow.com/questions/50634447/multiple-tasks-in-the-same-node-with-slurm

https://stackoverflow.com/questions/35498763/parallel-but-different-slurm-srun-job-step-invocations-not-working