nvcrectl cluster | NVIDIA Cluster Readiness Engine

nvcrectl cluster info

Discovers GPU nodes in the cluster and reports the detected platform, GPU architecture, per-node GPU count, and network topology (rack/T1 leaf switch grouping).

A node’s GPU count is its allocatable nvidia.com/gpu. On a DRA-only GPU stack, where no node advertises that resource, the count is the number of full GPUs the node publishes in its gpu.nvidia.com ResourceSlices.

nvcrectl cluster info [flags]

The topology key is auto-detected per cloud platform:

Platform	Topology label
AWS	`topology.k8s.aws/network-node-layer-1`
GCP	`cloud.google.com/gce-topology-block`
Azure	`kubernetes.azure.com/ppg`

Use --topology-key to override for other environments.

Flags

Flag	Default	Description
`--output` / `-o`	`table`	Output format: `table`, `json`, or `yaml`
`--topology-key`	auto	Node label for topology grouping

Example

# Table output (default)
nvcrectl cluster info
# JSON output for scripting
nvcrectl cluster info -o json

Sample output

Platform:     aws
GPU:          NVIDIA H100 80GB HBM3 (h100, 8 GPUs/node)
Nodes:        16 ready
TOPOLOGY (topology.k8s.aws/network-node-layer-1):
  rack-a    8 nodes
  rack-b    8 nodes