Storage and backups¶

CARC provides two broad classes of storage:

• Enterprise storage is intended for safely retaining data over the long term. Home directories and Enterprise Project Storage are backed up.
• Scratch storage is temporary working space for data that is being actively processed or analyzed. Scratch storage is not backed up and should not be used as the only copy of important data.

For help restoring data from Enterprise Project Storage or a home directory, email help@carc.unm.edu.

Storage limits¶

Run the quotas command to list your current quota usage and the storage locations available to your account and projects.

Easley scratch expiration¶

Easley scratch is intended only for active work. Files that have not been accessed for 180 days are automatically deleted. This expiration policy applies to both personal and project storage on Easley scratch. Deleted scratch data cannot be restored from backup. Move data that must be retained to Enterprise Project Storage or another appropriate long-term storage location before the expiration period is reached.

Choosing a storage type¶

• Use your home directory for personal source code, scripts, documents, and other relatively small files that are difficult to reproduce.
• Use Enterprise Project Storage for project data that must be retained safely over the long term.
• Use CARC or Easley scratch for temporary data being actively processed or analyzed, especially data that can be regenerated. Move anything that must be retained to enterprise storage before deleting the scratch copy.
• For especially I/O-intensive work, a job can temporarily use the compute node's /tmp disk or /dev/shm. Both are cleared when the job ends, and /dev/shm consumes memory that would otherwise be available to the job.

Home, project, and scratch filesystems are shared resources. Poorly chosen storage or inefficient I/O can affect other users as well as your own work. For file-count quotas and related policy details, see Storage and compute usage policies.

Finding your project space path¶

The project scratch paths in the storage-limits table use {PI username} for the CARC username of the project's PI and {name} for the project space name. Use the quotas command to list your current quota usage and storage locations, and check ColdFront for the project's allocation status. If a newly approved path does not exist yet, provisioning may not have finished; email help@carc.unm.edu if you cannot find or access the space.

Example use of a node's hard drive¶

This Slurm script first copies an input file (large_input_data.dat) to the compute node, runs the calculation ("run_my_program"), and then copies all results back into the submission directory:

#!/bin/bash
#SBATCH --nodes=1
#SBATCH --ntasks=8
#SBATCH --time=1:00:00
#SBATCH --job-name=local_storage

# Define a directory on the node-local disk, create it once the job
# has started, move data there, then cd to it and run
TEMP_DIR=/tmp/$USER/$SLURM_JOB_ID
mkdir -p "$TEMP_DIR"
cp -r $SLURM_SUBMIT_DIR/large_input_data.dat "$TEMP_DIR"
cd $TEMP_DIR

# Now run my program
run_my_program

# The job has finished, so move data back to where it came from
cp -r $TEMP_DIR/* $SLURM_SUBMIT_DIR

# Finally clean up the temporary directory
rm -r $TEMP_DIR

Video walkthrough¶

Storage Systems — from the CARC video tutorials:

Originally migrated from UNM-CARC QuickBytes; paths, tiers, and retention reconciled 2026-09-01 against cluster state observed 2026-07-25 (CARC knowledge store). Spotted a problem? Open an issue or pull request.

Machine-readable versions of this page: Markdown twin · raw source on GitHub · llms.txt · llms-full.txt (whole site). See For AI agents.