Installing NVIDIA Mission Control Components in a Connected Environment

Install each NVIDIA Mission Control component using the steps for that component. All packages and container images are pulled directly from their upstream sources (NGC, public Helm repositories) during installation — no local registry or bundle is required.

Before you deploy Mission Control autonomous job recovery, complete User-Cluster Operator Placement. After you install Run:ai and before you submit user workloads, complete Configure Run:ai Placement After Installation. Connected and air-gapped deployments use the same placement requirements.

The Autonomous Recovery Engine (ARE) consists of two independently installed components: Autonomous Hardware Recovery (AHR) and Autonomous Job Recovery (AJR). Both must be installed to enable full ARE functionality. Install both components after Run:ai, as shown in the following sequence. This ensures that the Run:ai BCM node categories and the k8s-user kubectl context exist when the ARE installation wizards run.

Keycloak and NMC Launchpad are installed through the GitOps-managed workflow (Installing GitOps-Managed Components).

Note

Run:ai installs on the k8s-user cluster (k8s-system-user nodes), not the k8s-admin cluster. Ensure the k8s-system-user node category and the GPU worker node category are created in BCM before running the Run:ai installation wizard.

• Argo CD
  • Overview
  • Prerequisites
  • Install the NMC Ansible collection
  • Prepare the Argo CD values file
  • Install Argo CD
  • Verify
  • Next step
• GitOps-Managed Components
  • Overview
  • Prerequisites
  • Prepare the values file
  • Prepare the secrets file
• NetQ
  • What’s New in NetQ 5.2
  • Release Considerations
  • NetQ Kubernetes Setup
  • Download the NetQ install packages
  • Installing NetQ
  • NetQ NvLink Configuration
• Observability Stack
  • Configure storage and retention for Prometheus
  • Configure storage and retention for Loki
  • Enable Prometheus scrape endpoint in BCM
  • Configure Prometheus to scrape metrics from BCM endpoint
  • Configure Prometheus to scrape metrics from DCGM endpoints
  • Grafana and querying BCM metrics
  • Forward Run:ai Metrics to the BCM Prometheus Exporter (Optional)
  • Install NMC Grafana dashboards
  • Reduce BCM monitoring directory size
  • Configure Grafana Single Sign-On with Keycloak
• Power Reservation Steering
  • Installation
  • Validation of PRS Installation
• NVIDIA Run:ai Installation
  • Run:ai Configuration
• NIM Operator
  • Installation
• NVIDIA Dynamo
  • Installation
  • Configuration
  • Node Categories
  • NVIDIA Run:ai Validation
• Autonomous Hardware Recovery (AHR)
  • Deployment Diagram
  • Prerequisites
  • Installation via BCM TUI Wizard
  • Login to the AHR UI
  • Installation Validation
  • Uninstallation via BCM TUI Wizard
  • AHR Appendix
• Autonomous Job Recovery (AJR)
  • Introduction
  • Prerequisites
  • User-Cluster Operator Placement
  • Before installing Mission Control autonomous job recovery
  • Configure TLS for Mission Control Autonomous Job Recovery Endpoints
  • Set up DNS resolution for Mission Control autonomous job recovery endpoints
  • Mission Control autonomous job recovery installation
  • Mission Control autonomous job recovery post-installation steps
  • Mission Control autonomous job recovery verification steps
  • Mission Control autonomous job recovery uninstallation
  • Third-Party Open Source Software Licenses
• Security Hardening
  • User and Admin Cluster Background
  • Securing Control Nodes
  • Restricting kubectl Access
  • Configuring Border Top Of Rack (BTOR) switching
  • Configuring Kyverno
  • Configuring PermissionManager