Archit Kulkarni

[Doc] [wip] Add dev workflow page

1

## Why are these changes needed? Adds a page describing a development workflow for Serve applications. ## Related issue number ## Checks - [ ] I've signed off every commit(by...

[Serve] [Doc] Refactor into new Scaling user guide, update code

3

## Why are these changes needed? Refactors the user guide to the following: Scaling and Resource Allocation - num_replicas - autoscaling - (fractional) resources Updates all code samples involved by...

## Why are these changes needed? ## Related issue number ## Checks - [ ] I've signed off every commit(by using the -s flag, i.e., `git commit -s`) in this...

[Core] [Dashboard] Allow `ray://` ("Ray Client") addresses for SubmissionClients

1

## Why are these changes needed? Users may have `RAY_ADDRESS` set to a Ray Client address (e.g. `ray://:10001`) in order to quickly connect to Ray. However, this is incompatible with...

## Why are these changes needed? TODO before merge: - [ ] Update doc - [ ] Add test ## Related issue number ## Checks - [ ] I've signed...

[Cluster launcher] Install latest Ray by default instead of nightly

9

## Why are these changes needed? In the default cluster launcher configs, previously the setup commands would install the *nightly* version of Ray if Ray was not already installed. This...

[RayService] [CI] Some tests for pending/active clusters may spuriously pass because head pod is not manually set to ready

2

The RayCluster will only switch if the new head Pod is ready, and the Ray Serve applications in the new RayCluster are prepared to serve requests. There is no Pod...

[Bug] worker group cannot be removed from RayCluster

2

### Search before asking - [X] I searched the [issues](https://github.com/ray-project/kuberay/issues) and found no similar issues. ### KubeRay Component ray-operator ### What happened + What you expected to happen Adding a...

[Doc] Add instructions to use `-aarch64` tag for images for M1 Mac in KubeRay YAMLs

2

Some users want to test KubeRay locally on M1 Mac, but it fails with unhelpful error messages, e.g. `Raylet is considered dead`. At a minimum, we should - Detect that...

The RayCluster health probe always uses the default value `52365`: https://github.com/ray-project/kuberay/blob/cbd8adbdd911da7f039fc912137a09e7b044478d/ray-operator/controllers/ray/common/pod.go#L260-L261 This will break if the user specifies a different `dashboard-agent-listen-port` in `RayStartParams`. _Originally posted by @architkulkarni in https://github.com/ray-project/kuberay/pull/1742#discussion_r1433231910_