volcano supports queue-level scheduling policies

[Monokaix]

What is the problem you're trying to solve

Volcano supports unified scheduling of online and offline workloads, provides a wealth of scheduling plugins and algorithms, and can distinguish different tenants through queue distinction. The current scheduling policy is a global configuration, and all jobs in the queue use the same scheduling policy, but in actual scenarios, different tenants may need to use different scheduling policies due to different usage scenarios. Therefore, volcano needs to support setting and using different scheduling policies at the queue level instead of using a globally unified scheduling policy.

Describe the solution you'd like

We can add a new field to the queue CRD, and users can set scheduling policies at the queue level, and volcano scheduler implements different scheduling policies based on the queue in which the job is located.

Additional context

No response

[RehanFazal77]

Hello @Monokaix , I’m really interested in contributing to this project as I have a good background in Kubernetes. I think the ability to set different scheduling policies at the queue level is an important improvement, and I’d love to be involved in implementing this.

Could you let me know how I can start contributing, and if there are any steps or tests to go through for the upcoming mentorship program? I’m eager to collaborate!

Thank you!

[hwdef]

This feature is very useful. For example, we need to deploy both online and offline jobs in a cluster. Some jobs require binpack, while others object to binpack.

[Monokaix]

This is a LFX project of CNCF, anyone who is interested can apply it here: https://mentorship.lfx.linuxfoundation.org/project/a785c059-fb70-41aa-88a2-62692ab2ca98

[hanshal101]

Hey @Monokaix and @zhoumingcheng , Hanshal here. I am an open-source developer, contributed to various organizations like BuildSafe, Glasskube, Cyclops, and a couple of Apache's. I have extensive experience in Golang and Kubernetes and am currently working at Cloudraft as a DevOps Intern. I'm planning to submit a project proposal for this issue in the LFX Term 1 2025. This feature looks pretty useful, as using different scheduling policies at the queue level instead of using a globally unified scheduling policy can solve a wide range of problems with different user's. Looking forward to be a part of Volcano Community too.

Thank You.