Cabinet Visualizer
Monitor the aggregate statistics of full cabinets with Grafana
The Cabinet Visualizer dashboard provides statistics and historical information about each cabinet, its cooling system, and the rack it contains. From here you can monitor the overall health of the cabinet and see important details about each Node in the enclosed rack to identify issues and track historical trends.
This dashboard is particularly useful for monitoring the overall health of GB200 NVL72-powered Node Pools and their constituent Nodes, as they are only available for deployment as full racks in dedicated cabinets.
Get started
To access the dashboard, first connect to CoreWeave's Managed Grafana instance from the Cloud Console, then select Dashboards.
Next, expand the Fleet Management section to reveal the Cabinet Visualizer dashboard.
Open the dashboard, then select the desired Cluster, Cluster org, Zone, and NVLink domain.
Dashboard overview
The Cabinet Visualizer dashboard is divided into several sections. The upper-left has aggregate statistics: GPU utilization, NVLink utilization, FP8 FLOP/s, and GPU temperature.
Below the aggregate statistics is a visual representation of the enclosed rack, with each Node identified by its name, and color-coded indicators for its NLCC (Node Life Cycle) state, Kubernetes state, and GPU temperature.
Hover over these indicators for more information, or click a Node to view its details.
To the right are time-series graphs showing the aggregate NVLink bandwidth and FP8 FLOP/s of this cabinet over time.
Below the time-series graphs, a table exposes more detail about the NLCC (Node Life Cycle) and Kubernetes states, the GPU and water cooling inlet temperatures, and system load. The tables are color-coded to indicate the health status of each Node.
By using the Cabinet Visualizer dashboard, you can monitor the health of your full cabinet, the enclosed rack, and its constituent Nodes to identify any issues that may arise.