rackslab/Slurm-web

What it solves

Slurm-web provides a graphical user interface for Slurm-based HPC clusters, replacing the need to rely solely on the command-line interface (CLI) for managing jobs and resources. It allows users and administrators to monitor supercomputers in a web browser across various devices.

How it works

It acts as a web dashboard that sits on top of Slurm, providing interactive charts, live status updates, and visualizations of node status and racking topology. It integrates with Prometheus for time-series metrics and supports various authentication methods like LDAP, Active Directory, and OIDC.

Who it’s for

Administrators and users of HPC supercomputers who need a more intuitive, visual way to track jobs, monitor GPU utilization, and manage cluster resources.

Highlights

• Resource Monitoring: Interactive charts for resource and job status, including specific monitoring for GPU utilization.
• Cluster Management: Support for multi-cluster environments and advanced RBAC permissions management.
• Visualizations: Advanced views of racking topology, account trees, QOS, and reservations.
• UTF-8 Support: Responsive web UI with dark mode and custom branding options.
• Enterprise Integration: LDAP, Active Directory, and OIDC authentication support.