Job container

Local temporary directories (/tmp, /dev/shm) are created on the festus computing nodes for each job. This means that these directories do not have to be cleaned up after the job has been executed, as this is done automatically. However, this also means that you cannot access /tmp and /dev/shm from outside the job, as you have to call up the job context to do this. To access such job container, which means access a node with a specific job context, you could use the command:

srun --jobid=<SLURM_JOBID> --overlap --nodelist=<NODE> <COMMAND>

Example

# srun --jobid=1234 --overlap -N 1 --ntasks-per-node=1 --nodelist=s73a01 ls /tmp

If you log on with ssh to a node on which you are running a job, you will not be able to access the same /tmp or /dev/shm as your job!