When you create a cluster, you typically provide the following values:
| Value | Description |
|---|---|
| Cluster name | Unique cluster name
|
| Description | Optional user-friendly description (128 character maximum length) |
| Node pool | Target AI Studio node pool for deployment |
| Worker scaling | Requested worker count, minimum 2 |
| Ray head pod resources | Resources for the Ray head node CPU, Memory, GPU |
| Ray work pod resources | Resources allocated per worker node: CPU, Memory, GPU |
For GPU workloads, select a GPU-capable node pool and set the worker GPU count and, if needed, the head GPU count. OPF-Ray automatically selects the GPU image, which includes CUDA-related libraries and GPU-focused data science packages.
Provisioning is asynchronous and may take a few minutes.