Filesystems Available

General information on NECluster file-systems

All nodes have system drives with the Linux OS installation. Data directories and NE-specific codes are shared across cluster from NFS servers via the shared Gigabit network.

Home directories

The home directories are exported from one NFS server. For this reason the input-output (I/O) performance may suffer if many users are accessing the same filesystem from multiple nodes. This can be alleviated by using other shared filesystems for I/O intensive programs, see below.

User data in are not backed up. Users are responsible for maintaining their own data backups.

Memory mounted /tmp/ directory

All nodes have /tmp/ mounted in memory for fast access. If your job runs on a single node, consider using that for temporary files. Create your directory /tmp/<username>/

Files are cleared on reboot.

Shared temporary directory in /home/tmp_scale/

SCALE runs write and read many temporary files. Please use /home/tmp_scale/ filesystem for these files, if you run SCALE in MPI mode across several nodes. This filesystem is exported from a separate NFS server, taking load off your /home/ directory. Create your own sub-directory /home/tmp_scale/<username>/ and point SCALE there.

Files are deleted from /home/tmp_scale/ after 180 days of last modification.

Shared scratch directory in /scratch/

Another separate NFS server exports /scratch/ directories. Please consider using /scratch/ for code development. Create your own sub-directory /scratch/<username>/

Files are deleted from /scratch/ after 180 days of last modification.

NE Cluster programs in /opt/ directory

Programs you want to run which are not a standard part of the Linux distribution (SCALE, MCNP, NESTLE, ...) are in /opt/ directory. Some of these codes require a license from RSICC, and you can only get access to them after providing a proof of obtaining the proper license.