Known issues and limitations for watsonx Code Assistant

The following known issues and limitations apply to the watsonx Code Assistant service.

Model scheduled on node with incompatible GPU type

Applies to: 5.2.2 and later

Problem

When deploying a foundation model, the model predictor pod might not start if it is scheduled on a node with an unsupported GPU type.

Cause

The cluster has GPU types that are not supported by the deployed foundation model.

Workaround

1. Run the following command to identify nodes that have GPU in your environment:

   oc get node -L nvidia.com/gpu.present

2. Verify that the GPU type on the node is supported by your foundation model. For more information, see GPU requirements for models
3. Ensure the model predictor pod runs on a node with a supported GPU type by applying your entitlements and, if necessary, pinning the pod to that node, see Applying your entitlements and pinning pods to nodes.