DRA: admin-controlled device attributes (device health, maintenance, priority)

[pohly]

Enhancement Description

• One-line enhancement description (can be used as a release note): By overriding device attributes, cluster admins can prevent usage of certain devices (unhealthy, taken offline for maintenance) or let the scheduler prefer certain devices over others (first use local devices, then devices connected via a fabric) with a Kubernetes API that works for all DRA drivers.
• Kubernetes Enhancement Proposal: KEP-5027: DRA: admin-controlled device attributes #5034
• Discussion Link: DRA: mark devices as unhealthy in ResourceSlice kubernetes#128979 (comment), DRA: structured parameters: network-attached resource kubernetes#124042 (comment)
• Primary contact (assignee): @pohly
• Responsible SIGs: SIG Node, SIG Scheduling
• Enhancement target (which target equals to which milestone):
  • Alpha release target (x.y): 1.33
  • Beta release target (x.y):
  • Stable release target (x.y):
• Alpha
  • KEP (k/enhancements) update PR(s):
  • Code (k/k) update PR(s):
  • Docs (k/website) update PR(s):

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

[pohly]

/sig node
/sig scheduling
/wg device-management

[eero-t]

By overriding device attributes, cluster admins can prevent usage of certain devices (unhealthy, taken offline for maintenance)

How admin should run test workload(s) on device which scheduling has been disabled (e.g. for firmware upgrade), to know whether it can be enabled egain (for production workloads)?

With node taints, one would use taint tolerance for this, but I don't seem from KEP description how similar thing would be achieved for DRA devices.

[pohly]

@eero-t: let's discuss in the PR, there we have threading => https://github.com/kubernetes/enhancements/pull/5034/files#r1913015929