[Feature] ray-service

[AlexAseyev]

Search before asking

• I had searched in the issues and found no similar feature requirement.

Description

I tested deploying RayServe and RayCluster to Kubernetes using Helm Chart and found some inconvenience.

How are new Served applications supposed to be deployed? I tried deploying:
https://raw.githubusercontent.com/ray-project/kuberay/master/ray-operator/config/samples/ray-service.stable-diffusion.yaml
and noticed that kind: RayService is used, unlike kind: RayCluster, which is specified in the helm-charts in this repo.

I think it makes sense to create a new helm-chart for RayServe here:
https://github.com/ray-project/kuberay/tree/master/helm-chart

How are new Served applications supposed to be added to an already deployed cluster?

What do you think? I think many would find it convenient to have a declarative Helm solution for GitOps. I might work on that.

Use case

No response

Related issues

No response

Are you willing to submit a PR?

• Yes I am willing to submit a PR!

[Future-Outlier]

Hi, @AlexAseyev
do you want to contribute?

[Future-Outlier]

Hi, @AlexAseyev

How are new Served applications supposed to be added to an already deployed cluster?

you can try this.
https://docs.ray.io/en/latest/serve/getting_started.html