KubeRay is a powerful, open-source Kubernetes operator that simplifies the deployment and management of Ray applications on Kubernetes. It offers several key components:
KubeRay core: This is the official, fully-maintained component of KubeRay that provides three custom resource definitions, RayCluster, RayJob, and RayService. These resources are designed to help you run a wide range of workloads with ease.
-
RayCluster: KubeRay fully manages the lifecycle of RayCluster, including cluster creation/deletion, autoscaling, and ensuring fault tolerance.
-
RayJob: With RayJob, KubeRay automatically creates a RayCluster and submits a job when the cluster is ready. You can also configure RayJob to automatically delete the RayCluster once the job finishes.
-
RayService: RayService is made up of two parts: a RayCluster and a Ray Serve deployment graph. RayService offers zero-downtime upgrades for RayCluster and high availability.
Community-managed components (optional): Some components are maintained by the KubeRay community.
-
KubeRay APIServer: It provides a layer of simplified configuration for KubeRay resources. The KubeRay API server is used internally by some organizations to back user interfaces for KubeRay resource management.
-
KubeRay Python client: This Python client library provides APIs to handle RayCluster from your Python application.
-
KubeRay CLI: KubeRay CLI provides the ability to manage KubeRay resources through command-line interface.
- A cloud-native, open-source stack for accelerating foundation model innovation IBM (May 9, 2023).
- AI/ML Models Batch Training at Scale with Open Data Hub Red Hat (May 15, 2023).
You can view detailed documentation and guides at https://ray-project.github.io/kuberay/.
We also recommend checking out the official Ray guides for deploying on Kubernetes at https://docs.ray.io/en/latest/cluster/kubernetes/index.html.
- Try this end-to-end example!
- Please choose the version you would like to install. The examples below use the latest stable version
v0.5.0.
| Version | Stable | Suggested Kubernetes Version |
|---|---|---|
| master | N | v1.19 - v1.25 |
| v0.5.0 | Y | v1.19 - v1.25 |
Make sure your Kubernetes and Kubectl versions are both within the suggested range. Once you have connected to a Kubernetes cluster, run the following commands to deploy the KubeRay Operator.
# case 1: kubectl >= v1.22.0
export KUBERAY_VERSION=v0.5.0
kubectl create -k "github.com/ray-project/kuberay/ray-operator/config/default?ref=${KUBERAY_VERSION}&timeout=90s"
# case 2: kubectl < v1.22.0
# Clone KubeRay repository and checkout to the desired branch e.g. `release-0.5`.
kubectl create -k ray-operator/config/defaultTo deploy both the KubeRay Operator and the optional KubeRay API Server run the following commands.
# case 1: kubectl >= v1.22.0
export KUBERAY_VERSION=v0.5.0
kubectl create -k "github.com/ray-project/kuberay/manifests/cluster-scope-resources?ref=${KUBERAY_VERSION}&timeout=90s"
kubectl apply -k "github.com/ray-project/kuberay/manifests/base?ref=${KUBERAY_VERSION}&timeout=90s"
# case 2: kubectl < v1.22.0
# Clone KubeRay repository and checkout to the desired branch e.g. `release-0.4`.
kubectl create -k manifests/cluster-scope-resources
kubectl apply -k manifests/baseObserve that we must use
kubectl createto install cluster-scoped resources. The correspondingkubectl applycommand will not work. See KubeRay issue #271.
A Helm chart is a collection of files that describe a related set of Kubernetes resources. It can help users to deploy the KubeRay Operator and Ray clusters conveniently. Please read kuberay-operator to deploy the operator and ray-cluster to deploy a configurable Ray cluster. To deploy the optional KubeRay API Server, see kuberay-apiserver.
helm repo add kuberay https://ray-project.github.io/kuberay-helm/
# Install both CRDs and KubeRay operator v0.5.0.
helm install kuberay-operator kuberay/kuberay-operator --version 0.5.0
# Check the KubeRay operator Pod in `default` namespace
kubectl get pods
# NAME READY STATUS RESTARTS AGE
# kuberay-operator-6fcbb94f64-mbfnr 1/1 Running 0 17sPlease read our CONTRIBUTING guide before making a pull request. Refer to our DEVELOPMENT to build and run tests locally.
Kuberay has an active community of developers. Here’s how to get involved with the Kuberay community:
Join our community: Join Ray community slack (search for Kuberay channel) or use our discussion board to ask questions and get answers.
If you discover a potential security issue in this project, or think you may have discovered a security issue, we ask that you notify KubeRay Security via our Slack Channel. Please do not create a public GitHub issue.
This project is licensed under the Apache-2.0 License.