What happened?

Resizing CPU requests gets stuck in InProgress if the CPU limit is not configured:

$ kubectl patch pod resize-pod --patch '{"spec":{"containers":[{"name":"resize-container", "resources":{"requests":{"cpu":"300m"}}}]}}'
pod/resize-pod patched
$ sleep 300
$ kubectl get pod resize-pod -o jsonpath='spec: {.spec.containers[0].resources}{"\nallocatedResources: "}{.status.containerStatuses[0].allocatedResources}{"\nstatus: "}{.status.containerStatuses[0].resources}{"\nresize: "}{.status.resize}{"\n"}'
spec: {"requests":{"cpu":"300m","memory":"200Mi"}}
allocatedResources: {"cpu":"300m","memory":"200Mi"}
status: {"requests":{"cpu":"200m","memory":"200Mi"}}
resize: InProgress

The internal behavior in kubelet varies depending on other resource limits configuration.

• If the limit for another resource is also not configured, no action is computed at computePodResizeAction():

  Jul 26 13:43:09 kind-control-plane kubelet[708]: I0726 13:43:09.494698     708 kuberuntime_manager.go:1051] "computePodActions got for pod" podActions="KillPod: false, CreateSandbox: false, UpdatePodResources: false, Attempt: 0, InitContainersToStart: [], ContainersToStart: [], EphemeralContainersToStart: [],ContainersToUpdate: map[], ContainersToKill: map[]" pod="default/resize-pod"
  

  This looks caused because container.Resources.Limits == nil:
  

  kubernetes/pkg/kubelet/kuberuntime/kuberuntime_manager.go

  Lines 555 to 557 in 3a8a60e

  	if container.Resources.Limits == nil || len(pod.Status.ContainerStatuses) == 0 {
  	return true
  	}

• If the limit for another resource is configured, doPodResizeAction() gets failed:

  Jul 26 14:04:23 kind-control-plane kubelet[708]: E0726 14:04:23.063108     708 kuberuntime_manager.go:750] "podResources.CPUQuota or podResources.CPUShares is nil" pod="resize-pod"
  

  This message is logged here:
  

  kubernetes/pkg/kubelet/kuberuntime/kuberuntime_manager.go

  Lines 754 to 757 in 3a8a60e

  	if podResources.CPUQuota == nil || podResources.CPUShares == nil {
  	klog.ErrorS(nil, "podResources.CPUQuota or podResources.CPUShares is nil", "pod", pod.Name)
  	result.Fail(fmt.Errorf("podResources.CPUQuota or podResources.CPUShares is nil for pod %s", pod.Name))
  	return

  
  This is caused because CPUQuota is not set if the CPU limit is not configured:
  

  kubernetes/pkg/kubelet/cm/helpers_linux.go

  Lines 179 to 182 in 3a8a60e

  	if cpuLimitsDeclared {
  	result.CPUQuota = &cpuQuota
  	result.CPUPeriod = &cpuPeriod
  	}

There is a similar problem in resizing memory limits. See the comment below: #126388 (comment)

What did you expect to happen?

This resizing should not get stuck in InProgress though I’m not sure whether this should be actuated or rejected yet.

How can we reproduce it (as minimally and precisely as possible)?

1. Enable InPlacePodVerticalScaling.

2. Create a pod where CPU request is configured and CPU limit is not configured:

   pod with no resource limits:

   apiVersion: v1
   kind: Pod
   metadata:
     creationTimestamp: null
     labels:
       run: resize-pod
     name: resize-pod
   spec:
     containers:
     - image: busybox
       name: resize-container
       command:
         - sh
         - -c
         - trap "exit 0" SIGTERM; while true; do sleep 1; done
       resources:
         requests:
           cpu: 200m
           memory: 200Mi
       resizePolicy:
       - resourceName: cpu
         restartPolicy: NotRequired
       - resourceName: memory
         restartPolicy: NotRequired
     restartPolicy: Always
   

   pod with memory limit:

   apiVersion: v1
   kind: Pod
   metadata:
     creationTimestamp: null
     labels:
       run: resize-pod
     name: resize-pod
   spec:
     containers:
     - image: busybox
       name: resize-container
       command:
         - sh
         - -c
         - trap "exit 0" SIGTERM; while true; do sleep 1; done
       resources:
         requests:
           cpu: 200m
           memory: 200Mi
         limits:
           memory: 200Mi
       resizePolicy:
       - resourceName: cpu
         restartPolicy: NotRequired
       - resourceName: memory
         restartPolicy: NotRequired
     restartPolicy: Always
   

3. Patch the pod to update its CPU request:

   $ kubectl patch pod resize-pod --patch '{"spec":{"containers":[{"name":"resize-container", "resources":{"requests":{"cpu":"300m"}}}]}}'
   

4. Wait for more than two minutes in order to confirm this issue is not the known issue.

5. See the pod resize status:

   $ kubectl get pod resize-pod -o jsonpath='spec: {.spec.containers[0].resources}{"\nallocatedResources: "}{.status.containerStatuses[0].allocatedResources}{"\nstatus: "}{.status.containerStatuses[0].resources}{"\nresize: "}{.status.resize}{"\n"}'
   spec: {"requests":{"cpu":"300m","memory":"200Mi"}}
   allocatedResources: {"cpu":"300m","memory":"200Mi"}
   status: {"requests":{"cpu":"200m","memory":"200Mi"}}
   resize: InProgress
   

Anything else we need to know?

No response

Kubernetes version

$ kubectl version
Client Version: v1.30.3
Kustomize Version: v5.0.4-0.20230601165947-6ce0bf390ce3
Server Version: v1.30.0

Cloud provider

N/A

OS version

# On Linux:
$ cat /etc/os-release
# paste output here
$ uname -a
# paste output here

# On Windows:
C:\> wmic os get Caption, Version, BuildNumber, OSArchitecture
# paste output here

Install tools

Container runtime (CRI) and version (if applicable)

Related plugins (CNI, CSI, ...) and versions (if applicable)