System Details on Zurada: How to Access and Use HPC System

Before You Begin

• Ensure you have a valid UofL login credentials.
• Review the Research Computing System Use Policies for compliance and security requirements.
• Request access through the Research Computing group if you are a new user.

Instructions

1. Request Access to Zurada
   Submit a service request ticket to the Research Computing group with a brief description of your research activity and resource needs (CPU, memory, GPU, duration).

2. Log In to Zurada
   Use your UofL credentials to log in to Zurada’s login nodes. Account sharing is prohibited; each user must use their own credentials.

3. Prepare and Submit Jobs

   • Place input/output files in your scratch directory. This directory is not backed up and files not accessed in 30 days may be deleted.
   • Use your home directory for code and small datasets (quota: 50 GB, backed up).
   • Submit jobs using the Slurm workload manager. Request only the resources needed for your job to ensure fair system use.
   • Avoid running computational jobs on login nodes.

4. Monitor and Manage Jobs

   • Each user is limited to a specific number of queued and active jobs.
   • Jobs have a maximum allowed run-time; use checkpointing/restart features if available.
   • Idle jobs or jobs not using allocated GPUs will be terminated after a 1-hour grace period.

5. Request Exceptions or Additional Resources

   • For longer job durations, more nodes or access to large-memory or GPU nodes, submit a service request ticket with your requirements and justification.

6. Review System Documentation and Policies

   • Consult system documentation for job limits, resource allocation, and best practices.
   • Follow all acceptable use policies; prohibited activities include account sharing, cryptocurrency mining and use of malicious software.

Further Readings

• UofL Research Computing Services

Need Additional Help

• Submit a Ticket: Use the self-service portal to log requests.