Table of contents
SLURM output and error files
SLURM job output and errors are written out to out%A.out and error%A.err files respectively, with %A being replaced by the job ID. The files are written respective to the directory where the job is submitted from, unless absolute path is specified in the SBATCH directive. Add the following in the sbatch job submission script:
#SBATCH -e error%A.err
#SBATCH -o out%A.out
In case of a job array, use the following format so that %a is replaced with the job step#.
#SBATCH -e error%A-%a.err
#SBATCH -o out%A-%a.out
/scratch
For fast file I/O performance, it is advisable to use the /scratch volume attached to the nodes.
# copying test_python.py from projects directory to /scratch directory
cp <your project directory>/test_python.py $SCRATCH
module load Python
python test_python.py
# copying output (results.txt) to the project directory
cp results.txt <your project directory>
rm -rf $SCRATCH