You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently we have the following per-pod VM customisations available
pod VM image
pod VM resources (cpu/mem/gpu)
pod VM instance type
These per-pod VM customisations are exposed as Kata hypervisor annotations in the pod spec. These annotations are available in the StartVM call for further processing before creating the pod VM.
There is a need to allow additional customisations like:
One of the suggestion from @mkulke was about using the concept of profiles to indicate the pod VM specification.
Conceptually this may involve the following:
Specification of pod VM profile: This will be specific for each provider. We can use configMaps for the profiles. Or we can use a CRD for the profile spec and use CRs for the profiles. Since we just want a structured mechanism to store profiles, this CRD can be without a controller.
Kata hypervisor annotation to specify the profile
CAA changes to query and use the profile for pod VM customisation
Currently we have the following per-pod VM customisations available
These per-pod VM customisations are exposed as Kata hypervisor annotations in the pod spec. These annotations are available in the StartVM call for further processing before creating the pod VM.
There is a need to allow additional customisations like:
We had some early discussions about this in the community slack channel (https://cloud-native.slack.com/archives/C04A2EJ70BX/p1734524383791669)
One of the suggestion from @mkulke was about using the concept of
profiles
to indicate the pod VM specification.Conceptually this may involve the following:
cc @mkulke @stevenhorsman
The text was updated successfully, but these errors were encountered: