Files
Dmitri GekhtmanandGitHub 7f833539fd Switch to 0.4.0 and eliminate Chart app versions. (#810)
Why are these changes needed?
This PR updates version tags in the release branch to 0.4.0.
Once this PR is committed to the release branch, final 0.4.0 artifacts will be built from the branch.

This PR should only be merged once we're ready to go through with the final release.

Removing app version
I've made a decision here to remove appVersions for the Helm charts.
appVersion is optional and I don't believe we need an extra version string floating around.
I'm open to keeping appVersion if anyone thinks it's needed.

Signed-off-by: Dmitri Gekhtman <[email protected]>
2022-12-08 11:22:01 -08:00
..
2021-12-21 10:02:23 +08:00
2022-09-06 02:16:25 -04:00

Ray Cluster

Make sure ray-operator has been deployed.

Ray is a unified framework for scaling AI and Python applications. Ray consists of a core distributed runtime and a toolkit of libraries (Ray AIR) for simplifying ML compute.

Helm

$ helm version
version.BuildInfo{Version:"v3.6.2", GitCommit:"ee407bdf364942bcb8e8c665f82e15aa28009b71", GitTreeState:"dirty", GoVersion:"go1.16.5"}

TL;DR;

# Because the ray-cluster chart in release 0.3.0 has some bugs, we need to clone the KubeRay repo and install the latest ray-cluster chart until release 0.4.0.
cd helm-chart/ray-cluster
helm install ray-cluster --namespace ray-system --create-namespace .

Installing the Chart

To install the chart with the release name my-release:

# Because the ray-cluster chart in release 0.3.0 has some bugs, we need to clone the KubeRay repo and install the latest ray-cluster chart until release 0.4.0.
cd helm-chart/ray-cluster
helm install my-release --namespace ray-system --create-namespace .

note: The chart will submit a RayCluster.

Uninstalling the Chart

To uninstall/delete the my-release deployment:

helm delete my-release -n ray-system

The command removes nearly all the Kubernetes components associated with the chart and deletes the release.

Check Cluster status

Get Service

$ kubectl get svc -l ray.io/cluster=ray-cluster
NAME                       TYPE        CLUSTER-IP     EXTERNAL-IP   PORT(S)                       AGE
ray-cluster-head-svc   ClusterIP   10.103.36.68   <none>        10001/TCP,6379/TCP,8265/TCP   9m24s

Forward to dashboard

$ kubectl get pod -o wide
NAME                                       READY   STATUS    RESTARTS   AGE    IP            NODE             NOMINATED NODE   READINESS GATES
ray-cluster-head-sd77l                 1/1     Running   0          8h     10.1.61.208   docker-desktop   <none>           <none>
ray-cluster-worker-workergroup-czxd6   1/1     Running   0          8h     10.1.61.207   docker-desktop   <none>           <none>
kuberay-operator-687785b964-jgfhv          1/1     Running   6          3d4h   10.1.61.196   docker-desktop   <none>           <none>

$ kubectl port-forward ray-cluster-head-sd77l 8265
Forwarding from 127.0.0.1:8265 -> 8265
Forwarding from [::1]:8265 -> 8265