Cabinet Wrangler - CoreWeave Docs

To view the dashboard, go to the Cabinet Wrangler dashboard.

For instructions about accessing CoreWeave Grafana dashboards, see Access and use CoreWeave Grafana dashboards.

The Cabinet Wrangler dashboard provides an overview of cabinet details and status, including GPU data and Node information across racks.

The dashboard contains the following panels.

Panel	Description
Any Production GPUs	Total count of nodes in racks with any number of healthy nodes.
NVL64 Production GPUs	Baseline count of nodes based on the number of racks with at least 16 healthy nodes (Rack Count * Threshold (16)).
NVL64+ Production GPUs	Total count of nodes in racks with at least 16 healthy nodes.
NVL72 Production GPUs	Total count of nodes in racks with 18 healthy nodes.
NVL64+ Active Production Percentage	Percentage of active production GPUs of type NVL64 or higher.
Production GPUs	Distribution of the classes of Production GPUs.
Number of Racks with Production Schedulable Nodes	Number of racks with schedulable Nodes for production workloads.
Number of Racks with Production Nodes (stacked)	Stacked chart of the data in the Number of Racks with Production Nodes panel.
GPU Utilization (max by `rack_name`)	Time-series graph of maximum GPU utilization aggregated by rack name.
Cabinet Details	Detailed table with Summary and NLCC State Counts tabs. Shows metrics such as rack name, Active status, Health rollup, and Kubernetes schedulable status.
Node Messages	Messages or logs related to the system’s Nodes.

Last modified on June 10, 2026