[Feature] Support History Server

[KunWuLuan]

Search before asking

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

Description

This issue aims to track multiple pull requests (PRs) related to implementing or enhancing support for a history server within the project.

Use case

No response

Related issues

No response

Are you willing to submit a PR?

• [x] Yes I am willing to submit a PR!

[KunWuLuan]

TODO:

• [ ] Design Doc
• [ ] Agent Implementation
• [ ] Front End Implementation
• [ ] Webserver Implementation
• [ ] Sidecar Injection By Kuberay

[liugs0213]

TODO:

• [ ] Design Doc[ ] Agent Implementation[ ] Front End Implementation[ ] Webserver Implementation[ ] Sidecar Injection By Kuberay

Hi, I’m also very interested in this feature. I noticed that there is already a KubeRay Dashboard. Do you plan to reuse/extend the existing KubeRay dashboard for the history server implementation, or build a new one from scratch?

[my-vegetable-has-exploded]

Here is a related issue #3884

[Myasuka]

@KunWuLuan I noticed that there already have a PR focusing on log collector, in my mind, this is just part of the whole plan. Thanks for the contribution and looks forward for the update of design docs.

[KunWuLuan]

https://github.com/ray-project/kuberay/pull/4241

https://github.com/ray-project/kuberay/pull/4242

https://github.com/ray-project/kuberay/pull/4187

All codes have been split into 3 PRs for reviewing. @Future-Outlier

[Future-Outlier]

Hi, @KunWuLuan thank you so much!! plz remove Chinese comment and I'll start review today

[Future-Outlier]

Hi, @chiayi will there be a chance to see your implementation about event server with this one https://github.com/ray-project/kuberay/pull/4241 ?

[metasyn]

Really looking forward to this landing. Thank you all for adding it.

[jmccarthy-lila]

Thanks for working on this feature, really excited about this as well. I am able to build and launch the historyserver from the Dockerfile and instructions on #4187 on kind, but selecting a cluster via /enter_cluster/... fails: could not read HTML file I can see in the local minio bucket the session logs - might be just a case of routing. Any suggestions for building and utilizing?

[Future-Outlier]

Hi @jmccarthy-lila, Here is a step by step guide: https://github.com/ray-project/kuberay/pull/4187#pullrequestreview-3509505276 I think it’s ok to wait until the alpha version is released, since it’s not done yet.

[my-vegetable-has-exploded]

Thanks for working on this feature, really excited about this as well. I am able to build and launch the historyserver from the Dockerfile and instructions on #4187 on kind, but selecting a cluster via /enter_cluster/... fails:感谢开发这个功能，我也对此感到非常兴奋。我能够按照 #4187 上的 Dockerfile 和说明在 kind 上构建并启动历史服务器，但通过 /enter_cluster/... 选择集群时失败了： could not read HTML file I can see in the local minio bucket the session logs - might be just a case of routing. Any suggestions for building and utilizing?我可以在本地 minio 存储桶中看到会话日志——可能只是路由问题。对于构建和使用有什么建议吗？

Maybe you should run npm ci && npm build firstly under historyserver/dashboard/v2.51.0/client to package the dashboard. Or just set BUILD_RAYSERVER_DASHBOARD to yes when build history server image.