DRA: Device Attributes in Downward API

[johnbelamaric]

Enhancement Description

For KubeVirt-based workloads that utilize devices, it is necessary to pass down device identifiers through to the workload. Those identifiers are available on the device objects, but are not otherwise easily found by the workloads themselves (though the DRA driver has access to them).

• One-line enhancement description (can be used as a release note): Device attributes in DRA devices can be published via the Pod's downward API
• Kubernetes Enhancement Proposal: https://github.com/kubernetes/enhancements/blob/master/keps/sig-node/5304-dra-attributes-downward-api/README.md
• Discussion Link:
  • https://docs.google.com/document/d/1qxI87VqGtgN7EAJlqVfxx86HGKEAc2A3SKru8nJHNkQ/edit?tab=t.0#bookmark=kix.jbfgd4128ogh
  • https://docs.google.com/document/d/1qxI87VqGtgN7EAJlqVfxx86HGKEAc2A3SKru8nJHNkQ/edit?tab=t.0#bookmark=id.o3cyftcoin10
• Primary contact (assignee): @alaypatel07
• Responsible SIGs: Node
• Enhancement target (which target equals to which milestone):
  • Alpha release target (x.y): 1.36
  • Beta release target (x.y):
  • Stable release target (x.y):
• Alpha
  • KEP (k/enhancements) update PR(s):
    • add kep for DRA Attributes Downward API #5606
  • Code (k/k) update PR(s):
    • KEP 5304: add implementation for discoverable DRA Device metadata in pod kubernetes#137086
  • Docs (k/website) update PR(s):
    • KEP-5304: add docs for downward API for device attribtues website#54716

Items for Beta:

• KEP-5304: DRA: add roundtrip test for device metadata API kubernetes#137839
• KEP-5304: DRA device metadata, add declarative validation for metadata types kubernetes#137761

Please keep this description up to date. This will help the Enhancement Team to track the evolution of the enhancement efficiently.

[johnbelamaric]

/wg device-management
/sig node
/cc @alaypatel07 @mortent @pohly @klueska

[johnbelamaric]

Aspirationally targeting 1.34 but may postpone.

[johnbelamaric]

There is a fair bit of complexity around how many attributes should be made available, and how to handle changes in attributes, disappearance of slices, pods that use existing resource claims, etc. @alaypatel07 will start a doc where we can discuss these and other aspects before writing the KEP.

[k8s-triage-robot]

The Kubernetes project currently lacks enough contributors to adequately respond to all issues.

This bot triages un-triaged issues according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue as fresh with /remove-lifecycle stale
• Close this issue with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle stale