What is wrong with my InPlacePodVerticalScaling configuration in Kubernetes?

DevOps Kubernetes
#1

I have a question about InPlacePodVerticalScaling.
I’m using Kubernetes 1.29, and has already enabled featuregate of InPlacePodVerticalScaling for all components:
kubelet/kube-controller-manager/kube-scheduler/kube-apiserver

But when I patch container like this, container is always restarted:
kubectl patch pod tonyaw-pod --patch ‘{“spec”:{“containers”:[{“name”:“test”,“resources”:{“requests”:{“cpu”:“800m”}}}]}}’
My resizePolicy is:
kubectl get pods tonyaw-pod -ojson | jq .spec.containers[0].resizePolicy [
{
“resourceName”: “cpu”,
“restartPolicy”: “NotRequired”
},
{
“resourceName”: “memory”,
“restartPolicy”: “RestartContainer”
}
]
Could you please help to check what I missed?
Thanks in advance!

#2

Ensure that your container runtime supports dynamic resource adjustments. As example: containerd versions below v1.6.9 do not have the CRI support needed for full end-to-end operation of this feature

#3

containerd --version
containerd http://containerd.io|containerd.io 1.6.21 3dce8eb055cbb6872793272b4f20ed16117344f8

#4

BTW, I can always see following message, thus I assume there is some wrong configuration in my env:
Normal Killing 5s kubelet Container test definition changed, will be restarted