Networking Minerva User Guide

Import Dashboards

Dashboards and their related Trends, Topology, Incidents, and Anomaly views are stored as JSON import/export sets in the Reference Configuration Package under minerva-configs/dashboards. Dashboards are split into two sets depending on tenant type: dashboards for admin tenants are located at dashboards/minerva-stack-monitoring, and dashboards for user tenants are located at dashboards/user-tenant. See Preconfigured Dashboards for more information. To export views and settings from a deployment, see Export Views and Settings.

Supported Dashboards

Stack monitoring (default tenant, admin use):

  • MinervaEvents

  • MinervaLag

  • MinervaPerformance

  • Daily Check Dashboard

General purpose (admin and user tenants):

  • MinervaIngestion

  • Telemetry Health

User tenant (notification-enabled views): import from minerva-configs/dashboards/user-tenant

Primary dashboards:

  • AI Cluster Health

  • Job Failure Root Cause Analysis

  • Interface Errors and State

  • Machine Learning Based Insights

  • NVLink Health

  • NVLink Interfaces

  • Traffic and Congestion

Supporting dashboards:

  • Break-Fix Deterministic Signals

  • GPU Summary by GPU

  • GPU Summary by Node

  • Log Insights

  • Performance Dashboard

  • Switch Physical Layer Errors

  • Professor-X

Folder Layout

Stack monitoring

File

Dashboards

Per-dashboard *_views_filtered.json

MinervaEvents, MinervaLag, MinervaPerformance, Daily Check, MinervaIngestion, Telemetry Health

COMBINED_admin-stack-monitoring_dashboards.json

Combined import for all stack monitoring dashboards

User tenant

Folder

Dashboards

user-tenant/cluster-health/

AI Cluster Health, Break-Fix Deterministic Signals

user-tenant/nvlink/

NVLink Health, NVLink Interfaces

user-tenant/professor-x/

Professor-X, Traffic and Congestion, Interface Errors and State, Per Switch Metrics and State

user-tenant/telemetry-health/

Telemetry Health, MinervaIngestion

user-tenant/ai-insights/

Machine Learning Based Insights, Log Insights

user-tenant/jobs-and-performance/

Job Failure RCA, Performance Dashboard

user-tenant/gpu/

GPU Summary by Node, GPU Summary by GPU

user-tenant/combined/

All basic user-tenant dashboards (combined import)

Import Dashboards

You can import any individual dashboard into whichever tenant you are logged in to. The bulk import automatically adjusts the username and tenant of imported views to the correct configuration.

Platform Admins, Tenant Admins, and Editors can import. Only Platform Admins can export.

For a new stack, use the combined JSON files to import the full basic dashboard set for each tenant in a single file.

  1. Make the current Reference Configuration Package available on a system that can access the NVIDIA Minerva UI.

  2. In the upper-right corner, click Settings menu and select Actions > Import.

  3. Click Select file and choose the JSON file for a dashboard or a combined set from the Reference Configuration Package.

  4. Click Import and check for any errors. When the import finishes, the page shows Import completed successfully. To stop an import in progress, click Stop.

Export Views and Settings

Platform Admins can export saved views, settings, and topology configuration to JSON files, for example to keep a copy or to move them to another deployment. An export includes every tenant, and private views too. User accounts and SSO settings aren't exported.

  1. Open the settings menu and select Actions > Export.

  2. Select Views, Settings, or Topology.

  3. Click Export in the row you want. To export everything in the list, use the first row: All Views, All Settings, or All Topology Config.

  4. Select All in one file, Each entry in a separate file, or both, and click Download.

The files are named minerva-export_<name>_<date and time>.json, with the time in UTC. If you choose separate files, your browser might ask permission to download several files.

To import the files, use Actions > Import as described above. Imported views are assigned to your tenant and user.

Last updated: