Templates

SLURM

#!/bin/bash -l

# File: slurm.tmpl
# Template for using clustermq against a SLURM backend

#SBATCH --job-name={{ job_name }}
#SBATCH --error={{ log_file | /dev/null }}
#SBATCH --mem-per-cpu={{ memory | 1024 }}
#SBATCH --array=1-{{ n_jobs }}
#SBATCH --cpus-per-task={{ cores | 1 }}


export OMP_NUM_THREADS={{ cores | 1 }}
ulimit -v $(( 1024 * {{ memory | 1024 }} ))
CMQ_AUTH={{ auth }} ${R_HOME}/bin/R --no-save --no-restore -e 'clustermq:::worker("{{ master }}")'

The main differences in this SLURM template compared to the one that comes as default with the package installation are

• explicit pointing to the correct R binary via the R_HOMEenvironment variable. This is very important for environments where there is multiple versions of R installed that are not necessarily accessible via the PATH environment variable.
• educated guess for nested parallelisation. i.e. if each job/task launched by clustermq can make use of OpenMP parallelisation (settings cores > 1) we aready set OMP_NUM_THREADS accordingly.

The template can be referenced either in .Rprofile or in the code itself via

options(
    clustermq.scheduler = "slurm",
    clustermq.template = "slurm.tmpl" 
)