Clusters

From ScorecWiki

Jump to: navigation, search

To use the SCOREC computing clusters you need to be added to the cluster's list of registered users - please send a request in to help@scorec.rpi.edu if you need access.

When you first ssh to a cluster's master node, your home directory there will be empty; your regular SCOREC home directory will be available at /import/users/username. Note that this is only available on the master node, not on the compute nodes.

Your home directory on each cluster (/users/username) is meant to hold your bash_profile and ssh keys only. Code and data should be stored on your bigtmp space - /fasttmp.

/bigtmp on the borg cluster is available on other SCOREC systems as /bigtmp/. Note that any files in your bigtmp directory stored outside of the Backup directory are not backed up, and cannot be considered safe for any long-term storage; the space is intended for intermediate data and simulation storage only.

Jobs on the clusters need to be run through (SLURM). Jobs found running outside of the queue will be killed. If you need to interactively work on the cluster, please use the salloc command, along with srun.

List of Clusters

See also

Personal tools