Rāpoi HPC Acceptable Use Policy¶

1. Overview¶

Rāpoi is a shared High-Performance Computing (HPC) resource provided to support research and teaching activities. All users are expected to use the system responsibly, efficiently, and with consideration for others. By accessing Rāpoi, you agree to comply with this policy, as well as all applicable University ICT and data governance policies.

2. Acceptable Use¶

2.1 Purpose of Use¶

Rāpoi resources are provided for academic research and teaching purposes only. The following are not permitted:

Commercial or for-profit work without explicit approval
Proxy work, including submitting jobs on behalf of others without authorisation
Sharing accounts or credentials

2.2 Fair and Responsible Use¶

Rāpoi is a shared system, and improper usage can negatively impact other users. A fair-share scheme is enforced to facilitate equitable access across all users and research groups.

Users must:

Use computational resources efficiently
Request only the resources required for their workloads
Follow Slurm scheduling policies and limits
Be familiar with and regularly review the Rāpoi documentation: https://vuw-research-computing.github.io/raapoi-docs/
Respond to queries from administrators, moderators, and other support staff in a timely manner

Users must not:

Attempt to subvert fair-share or other equitable access systems

Login nodes are shared access points used to interact with the cluster. They are intended for:

Submitting jobs to the scheduler, for example sbatch or srun
Editing scripts and files
Compiling small or short-running code
Small file transfers and data management
Monitoring job status

3.2 Correct Usage¶

All computational workloads must be submitted to compute nodes via the Slurm scheduler.
If interactive work with significant resources is required, users must request an interactive job, for example using srun or salloc.

Login nodes are not designed for computation or intensive I/O. The following activities are strictly prohibited on login node:

Running resource-intensive tasks such as CPU-, memory-, or I/O-heavy workloads
Executing code, apart from very short scripts
Launching large numbers of processes
Processing large datasets
Running workflows that resemble batch jobs
Running software that involves a graphical user interface (GUI), including but not limited to development environments such as VSCode and MATLAB, and visualisation or plotting tools such as ParaView. Such tools must be run on an appropriate partition, as outlined in the Rāpoi documentation. Such activities degrade performance for all users and may be terminated without notice.

4. Accounts and Security¶

Users are responsible for maintaining the security of their accounts:

Do not share passwords or SSH keys
Use secure access methods, such as SSH
Do not allow unauthorised access to the cluster
Report any suspected security issues immediately
Accounts are personal and non-transferable

5. Data and Storage¶

Rāpoi storage systems are intended for active research workflows, not for long-term archival use. Users must:

Keep only necessary data in the home directory
Regularly clean up temporary and intermediate files in Scratch and /tmp
Maintain external backups of important data
Avoid filling shared filesystems
Avoid installing local copies of software that is already installed through the module system
Data can be stored on the SoLAR system; inquiries should be directed to Digital Solutions: Research Services Failure to manage storage responsibly may impact other users.

6. Job Scheduling and Resource Use¶

All compute work must be carried out via the Slurm scheduler. Users must:

Submit jobs using appropriate resource requests, including CPU, memory, and time
Avoid oversubscription and underutilisation of resources
Test workflows with small jobs before running large-scale workloads
Use interactive jobs when required Improper job submissions may be restricted or terminated to maintain system stability.

7. Prohibited Activities¶

The following are considered misuse of the system:

Running compute jobs on the login node
Attempting to bypass the scheduler
Causing system instability, for example by filling storage or crashing services
Downloading or hosting non-research-related content
Using GUI tools on the login node, such as VSCode or MATLAB
Any activity that violates University policies or applicable laws

8. Enforcement and Access Control¶

At the discretion of the Rāpoi administrators, a first minor breach due to inexperience or genuine misunderstanding may be addressed through guidance rather than immediate loss of access, provided it does not cause significant disruption or risk. Violations of this policy may result in:

Immediate termination of running processes without prior warning
Limitations on compute resources
Temporary or permanent suspension of access
Reporting to supervisors or institutional authorities
Immediate revocation of access in cases of severe misuse

9. Account Lifecycle¶

Accounts that are unused for extended periods may be disabled or removed. Users must maintain valid affiliation with the University to retain access

10. Support¶

If you are unsure about correct usage or need help:

Consult the Rāpoi documentation: https://vuw-research-computing.github.io/raapoi-docs/
Contact the Rāpoi support team: Muhammad.hashmi@vuw.ac.nz