Frequently Asked Questions

Frequently Asked Questions#

This page answers common questions about using the NMT HPC cluster.

Account and Access#

How do I get an account on NMTHPC?#

See our Accounts and System Login page for detailed instructions on requesting an account.

I forgot my password. How do I reset it?#

Contact the HPC support team at hpc@nmthpc.atlassian.net to request a password reset. Include your username in the request.

Can I access NMTHPC from off campus?#

Yes, but you’ll need to connect through the NMT VPN first. Contact NMT IT Services for VPN access and configuration instructions.

How long do NMTHPC accounts remain active?#

Student accounts are reviewed annually. Faculty and staff accounts remain active as long as you’re affiliated with NMT. Inactive accounts may be deactivated after extended periods of non-use.

Storage and Data#

Where should I store my files?#

NMTHPC has two main filesystems. See Nodes and Filesystems for more details.

What are the storage quotas?#

Storage quotas vary. Check your current usage with:

```bash $ quota -s ```

How do I back up my data?#

NMTHPC filesystems are not automatically backed up. You’re responsible for backing up important data to other locations.

Running Jobs#

What’s the difference between login nodes and compute nodes?#

Login nodes are for:

  • Editing files

  • Compiling code

  • Submitting jobs

  • Managing files

Compute nodes are for:

  • Running simulations

  • Data analysis

  • Any computationally intensive work

Never run heavy computations on login nodes. Use SLURM to submit jobs to compute nodes.

How do I run a job?#

There are two main ways:

  1. Interactive jobs: For testing and development ```bash $ srun –pty bash ``` See Running Interactive Jobs

  2. Batch jobs: For production runs ```bash $ sbatch myjob.sh ``` See Running Batch Jobs

How do I check the status of my job?#

```bash $ squeue -u $USER ```

For more detailed information:

```bash $ scontrol show job JOBID ```

See Monitoring Resources for more commands.

Why is my job pending?#

Common reasons include:

  • Resources: The requested resources (CPUs, GPUs, memory) aren’t currently available

  • Priority: Other jobs have higher priority

  • Limits: You’ve reached your maximum number of running jobs

  • QOS: Quality of Service restrictions

Check with:

```bash $ squeue -u $USER ```

The `REASON` column shows why a job is pending.

How do I cancel a job?#

```bash $ scancel JOBID ```

To cancel all your jobs:

```bash $ scancel -u $USER ```

Software#

What software is available on NMTHPC?#

Use the `module` system to see available software:

```bash $ module avail ```

See Software Available on NMTHPC for a comprehensive list.

How do I load software?#

```bash $ module load softwarename ```

Example:

```bash $ module load python/3.11 ```

Can I install my own software?#

Yes! You can:

  • Install Python packages in your home directory with `pip install –user`

  • Create Anaconda environments

  • Compile software in your home directory

See specific software guides in the Software and Examples section.

I need software that’s not installed. What should I do?#

Contact HPC support at hpc@nmthpc.atlassian.net with:

  • Software name and version

  • Why you need it

  • Link to the software website

Best Practices#

How can I be a good NMTHPC user?#

  • Don’t run jobs on login nodes

  • Request only the resources you need

  • Test with short jobs before submitting long runs

  • Clean up old data you no longer need

  • Monitor your jobs and cancel failed ones

  • Acknowledge NMTHPC in publications

How should I structure my workflow?#

  1. Develop and test on your local machine when possible

  2. Transfer only necessary data to NMTHPC

  3. Test with small jobs or interactive sessions

  4. Scale up to full production runs

  5. Transfer results back to your local machine

  6. Clean up intermediate files

What are some common mistakes to avoid?#

  • Running computations on login nodes

  • Not specifying enough memory or time

  • Not checking job status after submission

  • Ignoring error messages

  • Requesting more resources than needed

Getting Help#

How do I contact HPC support?#

Email: hpc@nmthpc.atlassian.net

Include in your request:

  • Detailed description of the issue

  • Job IDs (if applicable)

  • Error messages

  • What you’ve already tried

What information should I include in a support request?#

  • For job issues: Job ID, error messages, SLURM script

  • For software issues: Software name, version, exact commands you ran

  • For login issues: Error messages, your operating system

  • For data transfer issues: Transfer method, file sizes, error messages