Ray Dashboard lacks built-in authentication, so use kubectl port-forward rather than LoadBalancer for access.
KubeRay Operator in ACK is in invitational preview. Submit a ticket to apply.
Prerequisites
Ensure that a Ray cluster is running in ACK:
-
A Ray cluster is running in ACK.
Access the Ray Dashboard
Metrics display in Ray Dashboard is disabled by default.
-
Get the Service name for the Ray cluster head pod in your namespace:
kubectl get svc -n ${RAY_CLUSTER_NS}Expected output:
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE myfirst-ray-cluster-head-svc ClusterIP 192.XX.XX.188 <none> 10001/TCP,8265/TCP,8080/TCP,6379/TCP,8000/TCP 18mNote the Service name (for example,
myfirst-ray-cluster-head-svc) for the next step. -
Forward port 8265 to the Ray cluster head Service in a separate terminal — this blocks the shell:
kubectl port-forward svc/myfirst-ray-cluster-head-svc --address 0.0.0.0 8265:8265 -n ${RAY_CLUSTER_NS}Expected output:
Forwarding from 0.0.0.0:8265 -> 8265 -
Open
http://127.0.0.1:8265/in your browser. Ray Dashboard displays the cluster overview.
Next steps
-
Submit a Ray job to run distributed tasks.
-
Autoscale Elastic Compute Service (ECS) nodes with the Ray and ACK autoscalers.
-
Autoscale Elastic Container Instance nodes with the Ray autoscaler.