Storage on the VSC

Check the VSC documentation for a description of how the storage at the KULeuven VSC is organized.

$VSC_HOME is your home folder, it’s where you end up when you log on to the VSC. It’s not meant to store data, actually, you shouldn’t do anything here. The path to this folder is /user/leuven/ABC/vscABCDE (ABCDE being the 5 digits of your VSC number) but you can use $VSC_HOME as a shortcut.

$VSC_DATA is you data folder, where you store data and scripts that you want to keep. You can access it from any VSC site. The actual path is /data/leuven/ABC/vscABCDE but you can use $VSC_DATA as a shortcut.

$VSC_SCRATCH is your scratch folder. It’s typically where you let the tools that you run store their results. Any results that you want keep are then copied to $VSC_DATA. It’s not shared between VSC sites and it is cleaned up regularly: any files older than 30 days are removed. The actual path is /scratch/leuven/ABC/vscABCDE but you can use $VSC_SCRATCH as a shortcut.

Because it is cleaned regularly it is important that you use cp to transfer files to your scratch folder and not mv since cp updates the time stamp of the file while mv does not (risking to generate files older than 30 days).

$VSC_SCRATCH_NODE is the scratch folder of the node that you’re working on. You can see it as a tmp folder where you let the tools store their results (even more so than your scratch folder). Any results that you want keep should be copied to $VSC_DATA in your script because it’s cleaned up immediately after the job has finished. The actual path is /local_scratch or /tmp on the node but you can use $VSC_SCRATCH_NODE as a shortcut.

On top of this there are also staging folders. They are big chunks of disk space that are attributed to a project when your personal scratch is not big enough. Since the data for the ONT trainings is huge we use a staging folder for the trainings. You can access it via /staging/leuven/stg_00143. The staging folders are not cleaned but they are also not backed up.