Stay on top of the cluster’s health with two complementary dashboards.

Dashboards at a glance

  • Simple status (cluster-status.ds.uchicago.edu): a lightweight view that tells you if the scheduler, login nodes, and key services are up.
  • Grafana metrics (graf.ds.uchicago.edu): detailed node-by-node charts for GPU/CPU usage, job pressure, historical utilization, and more.

Both update in real time and are public, so you can keep them open on any device while a job runs.

Simple status dashboard Grafana metrics

When to use each

  • Quick yes/no: Check the simple status page before filing a ticket or starting a workshop. It answers “is the cluster generally healthy?” in a single glance.
  • Deep dive: Use Grafana while debugging hung jobs, capacity questions, or GPU contention. You can drill into specific nodes, partitions, or GRES devices to see if what you’re experiencing matches cluster-wide activity.

Tips for incident triage

  1. Keep the simple status page open in a browser tab; it auto-refreshes and is mobile-friendly.
  2. Bookmark favorite Grafana dashboards (GPU saturation, node availability, Lustre/NFS throughput) so you can jump straight to the metrics that matter for your workflow.
  3. When reporting an issue, include timestamps and screenshots/links from either dashboard—this helps staff correlate your report with backend logs.