All of the Slurm processes we deploy in the pipeline creates Slurm standard out by default. We prefer to have this file stored in a separate directory named slurmlogs . i.e. #SBATCH --output slurmlogs/%j.out Any platform using an older version of Slurm ( < 23) can not create that slurmlogs directory during runtime. In other words, it has to be created before hand and not having the directory means the job will fail within the first second without any log, Therefore, it is ideal to mkdir -p set-x/slurmlogs during split_configfiles_to_sets.sh execution
All of the Slurm processes we deploy in the pipeline creates Slurm standard out by default. We prefer to have this file stored in a separate directory named
slurmlogs. i.e.#SBATCH --output slurmlogs/%j.outAny platform using an older version of Slurm ( < 23) can not create thatslurmlogsdirectory during runtime. In other words, it has to be created before hand and not having the directory means the job will fail within the first second without any log, Therefore, it is ideal tomkdir -p set-x/slurmlogsduringsplit_configfiles_to_sets.shexecution