-
Notifications
You must be signed in to change notification settings - Fork 700
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Kubernetes versions to be supported in the next training operator release #1803
Comments
I prefer |
No objection but if it's low effort to keep 1.24 support, it would be great. |
If we stop supporting 1.24, we can add CEL validation (#1708). I believe that adding validations makes increase UX. |
If that's the only benefit, I'd suggest holding off on dropping 1.24 support since many production systems are still running on that. |
It sounds good to me, and I'm ok with keeping to support 1.24. But probably, we can't support the scheduler-plugins integration in v1.24 because of #1769. In the next release, we support essential features of the training operator in v1.24-v1.26, although we don't support the scheduler plugin integration in v1.24. |
This means users can upgrade the training operator to the next one in K8s v1.24 only if they don't use the scheduler-plugins integration. If they want to use the latest training operator with the scheduler plugins integration, they must use K8s |
Ok I am fine with dropping v1.24 support. |
Thank you! |
IMO, supporting 1.24 would be tough. I am proposing to keep 1.25 as the minimum k8s version. The community is proposing for supporting till 1.27 for the next KF release |
Thank you for letting us know! |
We can support 1.25, 1.26 and 1.27 in v1.7.0 release for being consistent |
/assign In the first step, I'll try to bump k8s version to v1.26. |
I will create a PR to upgrade k8s to v1.27 after #1769 is done. |
I found that k8s v1.24 is blocking other tasks. |
So I will upgrade k8s version, first. |
Currently, we support 3 K8s versions, such as v1.23, v1.24, and v1.25.
However, v1.23 has already reached EoL, and v1.24 will reach EoL on 2023-07-28.
So we can support
v1.24-v1.26
orv1.25-v1.27
in the next training operator release.ref: https://kubernetes.io/releases/
WDYT? @kubeflow/wg-training-leads @zw0610 @kuizhiqing
The text was updated successfully, but these errors were encountered: