-
Notifications
You must be signed in to change notification settings - Fork 312
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
[proposal] numa topology policy on pod #1912
Labels
Milestone
Comments
3 tasks
/close |
@jasonliu747: Closing this issue. In response to this:
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
What is your proposal:
Introduce new api for pod level numa topology policy. With the new api, users can specify numa topology policy for each
pod, so that pods that are more sensitive to latency can decide how they need to be orchestrated, rather than being passively
scheduled according to the numa topology policy on the node.
Why is this needed:
With numa topology policy set on node, users need to set nodeSelector or nodeAffinity to place the application under the
same numa. This means to split the nodes in cluster into static groups. In large clusters, with lots of nodes that can serve
different purposes, it's not unreasonable to dedicate a node (or set of nodes) to a certain numa topology policy, and
direct pods to those nodes using a nodeSelector. It's mostly problematic in smaller clusters where you don't have the
luxury of special-purposing nodes like this and you want to binpack pods onto nodes as tightly as possible (while still
reaping the benefits of topology-alignment). So we need a way to specify numa topology policy for each pod.
Is there a suggested solution, if so, please add it:
#1910
The text was updated successfully, but these errors were encountered: