Get available metrics | NVIDIA Fleet Intelligence Documentation

Get a list of all available metrics, organized by component. Supports optional filtering by isCritical, aggregatable, and type. Unknown or false values for boolean params are treated as no filter. Use psirtComponent=true to include active PSIRT CVE sub-components under the “psirt” component.

Query parameters

isCriticalstringOptional

If 'true', only critical metrics are returned. If absent or 'false', all metrics are returned.

aggregatablestringOptional

If 'true', metrics that cannot be aggregated across nodes are excluded. If absent or 'false', all metrics are returned.

typestringOptional

If set, only metrics matching this type are returned (e.g., ‘utilization’, ‘limit’). If absent, all metrics are returned.

dataSourceenumOptional

Filter the catalog by data source: inband, oob, or all. Omitted defaults to all sources.

Allowed values:

psirtComponentbooleanOptional

Include active PSIRT CVE sub-components (default: false)

entityLevelenumOptional

Optional entity scope. Components an entity cannot produce are omitted, so an empty result is the signal that a scope has no nodes of the requested data source. ‘node’ additionally filters by GPU/CPU architecture and probes per-sensor availability.

Allowed values:

entityIDstringOptional

Entity ID matching entityLevel: node UUID, node group UUID, or compute zone UUID. Required when entityLevel is set.

Response

Successful response

componentslist of objectsOptional

Errors

401

Unauthorized Error

403

Forbidden Error

499

Client Closed Request Error

500

Internal Server Error