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
Then the pod stays in ContainerCreating, and checking events show
Failed to create pod sandbox: rpc error: code = Unknown desc = failed to set up sandbox container "7fac6d98a43b291ebb2b9b0386cc4bb5e14c5b78984c27d6dd1ba438907990b6" network for pod "windows-runtimeclass": networkPlugin cni failed to set up pod "windows-runtimeclass_vlatombe" network: failed to parse Kubernetes args: pod does not have label vpc.amazonaws.com/PrivateIPv4Address
Inspecting the resulting pod spec, the resources requests and limits vpc.amazonaws.com/PrivateIPv4Address are unset.
Why issue was resolved without needing changes in VPC Resource Controller
The issue was resolved with the change to automatically merge the RuntimeClass's nodeSelector is into the pod's nodeSelector in admission, thereby meaning the VPC resource controller is always aware of the pods scheduling needs and can therefore handle IP address management on behalf of the pod. This is handled by the in-tree RuntimeClass admission controller.
I suspect that at the time the issue was raised, this mechanism may have been disabled behind an alpha feature gate which therefore required the workaround mentioned. The Runtimeclass admission controller feature went GA in 1.20 release. See references for more.
Community Note
Tell us about your request
Kubernetes 1.17 and above provides RuntimeClass, which simplifies the process of using taints and tolerations.
Which service(s) is this request for?
EKS
Tell us about the problem you're trying to solve. What are you trying to do, and why is it hard?
I set up a
RuntimeClass
as followthen I schedule a pod referencing the
RuntimeClass
Then the pod stays in
ContainerCreating
, and checking events showInspecting the resulting pod spec, the resources requests and limits
vpc.amazonaws.com/PrivateIPv4Address
are unset.Are you currently working around this issue?
Adding to the container
allows the pod to be scheduled.
The text was updated successfully, but these errors were encountered: