Cabinet Wrangler
Monitor cabinet metrics with Cabinet Wrangler
Info
For accessing CoreWeave Grafana Dashboards instructions, see Access CoreWeave Grafana Dashboards.
The Cabinet Wrangler dashboard provides an overview of the cabinet details and status, including GPU data and Node information across racks.
The dashboard contains the following panels.
Panel | Description |
---|---|
Schedulable Production GPUs (Any) | Displays a count of all schedulable production GPUs, regardless of type. |
Schedulable Production GPUs (NVL64+) | Displays a count of schedulable production GPUs of type NVL64 or higher. |
Schedulable Production GPUs (NVL64) | Displays a count of schedulable production GPUs of type NVL64. |
Schedulable Production GPUs (NVL72) | Displays a count of schedulable production GPUs of type NVL72. |
Active Production (NVL64+) | Displays a count of active production GPUs of type NVL64 or higher. |
Number of Racks with Production Schedulable Nodes | A panel intended to show the number of racks that have schedulable Nodes for production workloads. |
Cabinet Details | A detailed table with "Summary" and "NLCC State Counts" tabs. It's designed to show metrics like NVlink domain, Active status, Health rollup, K8s schedulable status, and more. |
Node Messages | A panel meant to display messages or logs related to the system's Nodes. |