Skip to navigation
Platform ManagementSettings

General Settings

View as Markdown

The General settings section in the Admin panel provides centralized controls for enabling or disabling key platform features. These settings allow administrators to tailor the behavior of the NVIDIA Run:ai environment based on organizational policies, user needs, and workload types.

Permissions Required

The General Settings page is visible only when the user has at least one settings permission with CRUD (Create, Read, Update, Delete) access. If all of the user’s settings permissions are read-only, the page is not visible. To grant or modify these permissions, see Roles.

Analytics

SettingDescription
GPU profiling metricsEnable to view metrics in the Nodes table and the Workload Metrics tab. To configure the NVIDIA exporter and NVIDIA Run:ai cluster, see GPU profiling metrics

Resources

SettingDescription
Limit projects from exceeding department quotaEnable to prevent projects’ total quota from exceeding the department’s quota or being less than the total usage by non-preemptible workloads. For more information, see the NVIDIA Run:ai Scheduler guide.
CPU quotaEnable resource quota based on CPU resources. For more information, see the Departments guide and the Projects guide.
Over quota weightEnable over quota weight to set what proportion of unused resources a project can get on top of its quota. For more information, see the Over quota weight guide.
GPU resource optimization BetaEnable to set a GPU memory limit for workloads. To enable this on each NVIDIA Run:ai cluster, see the Dynamic GPU fractions guide.

Workloads

SettingDescription
PoliciesEnable to impose restrictions and default values on workloads’ fields. For more information, see the Policies guide
Hugging face modelsEnable instant deployment of Hugging Face models via the Models view or the workload submission form. For more information, see the Deploy inference workloads from Hugging Face.
NIM modelsEnable instant deployment of NIM models via the workload submission form. For more information, see the Deploy inference workloads with NVIDIA NIM.
Flexible workload submissionEnable to submit workloads via the new flexible form. Allow selecting an existing setup or starting from scratch, reviewing existing setups and understanding policy definitions. The form also includes advanced features tailored for inference workloads. For more information, see Submission form options.
Flexible workload templatesEnable to create workload templates via the new flexible form. For more information, see the Workload Templates guide.
Data volumesEnable to allow creation of data volumes to share underlying data across scopes. For more information, see the Data volumes guide.
Submit supported workload types via YAMLEnable to submit a range of workload types using a standard Kubernetes YAML. For more information, see the Submit supported workload types via YAML guide.
AI application submission ExperimentalEnable to deploy agentic AI application stacks from the NGC catalog or Helm repositories, with resource overrides for multi-component pipelines. For more information, see the AI applications guide.
NGC catalog BetaEnable to browse and select container images from the NVIDIA NGC catalog for faster environment setup.
NGC private registry BetaEnable browsing of your organization’s proprietary container image registry, hosted on the NGC private registry service

Security

SettingDescription
Identity providerAnyone given access to the NVIDIA Run:ai application via the identity provider system will be able to join the platform and authenticate using SSO. See Single Sign-On (SSO) guide.
Set logout redirect URLUsers will be redirected to this URL after logging out from NVIDIA Run:ai
Set UI inactivity timeoutUsers will be automatically logged out of NVIDIA Run:ai after the set period of inactivity. CLI and API access are not affected. (If left blank, users are logged out after 24 hours of inactivity by default.)

Notifications

SettingDescription
Email notificationsSet the email server through which the email notifications will be sent
Slack notificationsAuthorize and connect your Slack workspace to NVIDIA Run:ai
System notificationWrite a message that will be displayed to all users

For more details, see Notifications.

Cluster Authentication

To allow users to securely submit workloads using kubectl, you must configure the Kubernetes API server to authenticate users via the NVIDIA Run:ai identity provider. This is done by adding OpenID Connect (OIDC) flags to the Kubernetes API server configuration on each cluster. See Cluster authentication for more details.

Branding

Upload the logo you want displayed in the top-right corner of the NVIDIA Run:ai platform interface. Logos can be uploaded in SVG or PNG format (up to 128 KB).

To display a custom logo in the platform interface:

  1. Click + LOGO.
  2. In the upload window, click the upload icon and select your logo file
  3. Click SAVE

You can update or remove it at any time by clicking the edit or remove icons.

Using MCP

The NVIDIA Run:ai MCP server exposes Cluster and Infrastructure tools to manage tenant-wide feature flags via natural language:

  • list_system_settings - list tenant-wide platform feature flags and their current values
  • update_system_settings - toggle a tenant-wide platform feature flag on or off

See AI Agents for setup.