tests/ai-conformance: swap order of ray and kueue

Author: justinsb

tests/e2e/scenarios/ai-conformance/run-test.sh

@@ -98,6 +98,8 @@ spec:
 EOF
 
 ${KOPS} update cluster --name "${CLUSTER_NAME}" --yes --admin
+
+# TODO: Can we delay this until later?
 ${KOPS} validate cluster --wait=10m
 
 echo "----------------------------------------------------------------"
@@ -129,15 +131,6 @@ helm upgrade -i nvidia-gpu-operator --wait \
 # Uses the driver installed by GPU Operator at /run/nvidia/driver (the default).
 echo "Installing NVIDIA DRA Driver..."
 
-cat > values.yaml <<EOF
-# The driver daemonset needs a toleration for the nvidia.com/gpu taint
-kubeletPlugin:
-  tolerations:
-  - key: nvidia.com/gpu
-    operator: Exists
-    effect: NoSchedule
-EOF
-
 helm upgrade -i nvidia-dra-driver-gpu nvidia/nvidia-dra-driver-gpu \
     --version="25.12.0" \
     --create-namespace \
@@ -157,10 +150,9 @@ kubectl apply --server-side -f https://github.com/kubernetes-sigs/kueue/releases
 # Wait for webhooks to be ready (so we can apply other objects)
 kubectl rollout status deployment -n kueue-system kueue-controller-manager --timeout=5m
 
-# 3. Robust Controller (KubeRay)
+# KubeRay
 echo "Installing KubeRay Operator..."
-# KubeRay 1.3.0
-kubectl apply -k "github.com/ray-project/kuberay/ray-operator/config/default?ref=v1.5.0"
+kubectl apply --server-side -k "github.com/ray-project/kuberay/ray-operator/config/default-with-webhooks?ref=v1.5.0"
 
 echo "----------------------------------------------------------------"
 echo "Verifying Cluster and Components"