Observability Guide

NVIDIA Enterprise Reference Architecture

• Abstract
• Overview of the Observability Guide for NVIDIA Enterprise Reference Architectures (Enterprise RAs)
• Technical Prerequisites for Observability
  • NVIDIA Enterprise RAs – Prerequisites
  • Verifying Kube Prometheus and Grafana Setup
  • Configuring Persistent Storage for Grafana
  • Steps to configure a PVC for Grafana
  • Network Monitoring with NVIDIA NetQ
• Configuration of Data Sources
  • Base Command Manager – Prometheus
  • DCGM Exporter
  • NIM Metrics – Kube Prometheus
  • NetQ Prometheus
• Custom Dashboards
  • GPU Dashboard
  • NIMs Dashboard
  • Kubernetes Global Dashboard
  • CPU & Node Storage Dashboard
  • Network Dashboard
• Alerting
  • Configuring Grafana for Alerting
    • Folder Creation
    • Contact Point Creation
    • Setting up SMTP Server for configuration
  • Automated Alert Uploads
    • Script Overview
  • Alert Rule JSONs
• Summary
• Out of Scope
• Appendix A

Notices

• NVIDIA Enterprise Reference Architecture Notice & Disclaimers