Story

As an upstream contributor, I'd like to get the Device Plugin serial tests fixed, so that we can expect useful test results from the whole suite.

Failing tests as time of writing

• [It] [sig-node] Device Plugin [Serial] [Feature:DevicePlugin] DevicePlugin [Serial] [Disruptive] Can schedule a pod with a restartable init container [Feature:SidecarContainers]
• [It] [sig-node] Device Plugin [Serial] [Feature:DevicePlugin] DevicePlugin [Serial] [Disruptive] Does not keep device plugin assignments across node reboots if fails admission (no pod restart, no device plugin re-registration)
• [It] [sig-node] Device Plugin [Serial] [Feature:DevicePlugin] DevicePlugin [Serial] [Disruptive] Keeps device plugin assignments after the device plugin has restarted (no kubelet restart, pod restart)
• [It] [sig-node] Device Plugin [Serial] [Feature:DevicePlugin] DevicePlugin [Serial] [Disruptive] can make a CDI device accessible in a container
• [It] [sig-node] Device Plugin [Serial] [Feature:DevicePlugin] DevicePlugin [Serial] [Disruptive] Can schedule a pod that requires a device
• [It] [sig-node] Device Plugin [Serial] [Feature:DevicePlugin] DevicePlugin [Serial] [Disruptive] Keeps device plugin assignments across kubelet restarts (no pod restart, no device plugin restart)
• [It] [sig-node] Device Plugin [Serial] [Feature:DevicePlugin] DevicePlugin [Serial] [Disruptive] Keeps device plugin assignments after kubelet restart and device plugin restart (no pod restart)
• [It] [sig-node] Device Plugin [Serial] [Feature:DevicePlugin] DevicePlugin [Serial] [Disruptive] [OrphanedPods] Ensures pods consuming devices deleted while kubelet is down are cleaned up correctly

Testgrid Dashboards

• https://testgrid.k8s.io/sig-node-cri-o#node-kubelet-cgroupv1-serial-crio
• https://testgrid.k8s.io/sig-node-cri-o#node-kubelet-cgroupv2-serial-crio