Home / BeaverDeck / Docs / Insights Guide / GPU Insights / GPU Capacity Discovery

GPU Capacity Discovery

BeaverDeck uses this check to identify a specific gpu condition that may need operator review.

Permissions: viewing checks requires insights: view. Opening a linked object or logs requires the corresponding resource permission, and the BeaverDeck ServiceAccount must be allowed to read the Kubernetes resources used by the check. Suppressing a finding requires insights: edit and affects all users.

Check type	`gpu-capacity-discovery`
Insights section	GPU Insights
Alert severity	Critical

When It Reports A Finding

Selected namespaces contain active GPU requests, but no Node advertises allocatable nvidia.com/gpu capacity.

Why This Is A Problem

GPU workloads cannot be scheduled when the cluster exposes no compatible GPU resource, even if GPU hardware physically exists.

Recommended Response

Confirm that the expected GPU nodes are registered and Ready.
Check the NVIDIA device plugin, drivers, runtime configuration, and node allocatable resources.
Verify that workload resource names and cluster GPU resource names match.
Add or repair GPU capacity, then confirm that Nodes advertise nvidia.com/gpu.

Scope And Limitations

The check uses Kubernetes-advertised capacity. It cannot detect physical GPUs that are not exposed by the device plugin. When neither GPU workloads nor GPU nodes exist, the check reports a passing informational state.

After remediation: refresh GPU Insights and verify the underlying resource or metric. Suppress the finding only when the condition is intentional and its risk is accepted.