Operator Guide

​

What Operators Commonly Do

• Trigger or observe enrolment and validation workflows
• Check device state in NetBox before and after workflow execution
• Inspect Temporal workflow history when a job stalls or fails
• Use Grafana dashboards and Loki logs to understand system health
• Validate SLURM connectivity and burn-in execution for hardware-test workflows
• Manage tenants and allocations for customer-facing Radar projection

​

Systems Operators Regularly Inspect

• Fleet Manager API/UI for job submission and status
• Temporal for workflow history and failure context
• Grafana for logs, metrics, and dashboards
• NetBox for inventory and lifecycle-state validation

​

Start Here

• Use Getting Access for the underlying access prerequisites
• Use Status Reference for job and job-device status values
• Use DC Standup Pipeline to understand Operational Racks placement, statuses, gates, and actions
• Use Workflow State Transitions to understand how workflow runs affect device state and eligibility
• Use Device State Machines for all device status axes and state diagrams
• Use Delegated Device Health to interpret health status for delegated Metal devices
• Use Burn-in Testing Guide to run burn-in tests on GB300 hardware, from single-node through multi-pod
• Use Site Validation for the concrete validation checklist
• Use PKI & Certificate Trust when a site is being onboarded or certificates need to be understood
• Use Managing Tenants and Allocations before changing customer capacity assignments or Radar projection metadata
• Use Matrix Connector Operations to operate the new matrix-connector inventory pipeline — health checks, failure modes, and who to call