VMAgent and VPA: fails when .spec.shardCount is unset

Problem

Using victoria-metrics-k8s-stack and enabling vpa for vmagent still produces an error that ".spec.shardCount" does not exist.

Versions

victoria-metrics-k8s-stack:0.88.0
vm-operator: v0.74.0

VPA

apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  labels:
    app.kubernetes.io/component: monitoring
    app.kubernetes.io/instance: victoriametrics-eu-west1
    app.kubernetes.io/name: vmagent
    managed-by: vm-operator
  name: victoriametrics-eu-west1
  namespace: monitoring
  ownerReferences:
  - apiVersion: operator.victoriametrics.com/v1beta1
    blockOwnerDeletion: true
    controller: true
    kind: VMAgent
    name: victoriametrics-eu-west1
spec:
  resourcePolicy:
    containerPolicies:
    - containerName: '*'
  targetRef:
    apiVersion: operator.victoriametrics.com/v1beta1
    kind: VMAgent
    name: victoriametrics-eu-west1
  updatePolicy:
    updateMode: InPlaceOrRecreate
status:
  conditions:
  - lastTransitionTime: "2026-08-04T09:55:44Z"
    message: 'Cannot read targetRef. Reason: unhandled targetRef operator.victoriametrics.com/v1beta1
      / VMAgent / victoriametrics-eu-west1, last error Internal error occurred: the
      spec replicas field ".spec.shardCount" does not exist'
    status: "True"
    type: ConfigUnsupported
  - lastTransitionTime: "2026-08-04T09:55:44Z"
    message: 'No recommendation provided from the recommendation engine: No pods found.'
    reason: 'No recommendation provided from the recommendation engine: No pods found.'
    status: "False"
    type: RecommendationProvided
  recommendation: {}

VMAgent

apiVersion: operator.victoriametrics.com/v1beta1
kind: VMAgent
metadata:
  annotations:
    meta.helm.sh/release-name: victoriametrics
    meta.helm.sh/release-namespace: monitoring
  labels:
    app.kubernetes.io/component: vmagent
    app.kubernetes.io/instance: victoriametrics
    app.kubernetes.io/managed-by: Helm
    app.kubernetes.io/name: eu-west1
    app.kubernetes.io/version: v1.148.0
    helm.sh/chart: victoria-metrics-k8s-stack-0.88.0_216b0562e4f4
    helm.toolkit.fluxcd.io/name: victoriametrics
    helm.toolkit.fluxcd.io/namespace: monitoring
  name: victoriametrics-eu-west1
  namespace: monitoring
spec:
  externalLabels:
    cluster: eu-west1
  extraArgs:
    promscrape.streamParse: "true"
  image:
    tag: v1.148.0
  podDisruptionBudget:
    minAvailable: 1
  port: "8429"
  readinessProbe:
    initialDelaySeconds: 5
    periodSeconds: 10
    successThreshold: 1
    timeoutSeconds: 5
  remoteWrite:
  - url: http://vminsert-victoriametrics-eu-west4.monitoring.svc.clusterset.local:8480/insert/0/prometheus/api/v1/write
  - url: http://vminsert-victoriametrics-eu-west1.monitoring.svc.cluster.local:8480/insert/0/prometheus/api/v1/write
  replicaCount: 2
  resources:
    limits:
      memory: 700Mi
    requests:
      cpu: "0.2"
      memory: 700Mi
  scrapeInterval: 20s
  selectAllByDefault: true
  vpa:
    resourcePolicy:
      containerPolicies:
      - containerName: '*'
    updatePolicy:
      updateMode: InPlaceOrRecreate

.status.shards is 1

Description

so running kubectl get --raw /apis/operator.victoriametrics.com/v1beta1/namespaces/monitoring/vmagents/victoriametrics-eu-west1/scale gets
Error from server (InternalError): Internal error occurred: the spec replicas field ".spec.shardCount" does not exist

I have found issue https://github.com/VictoriaMetrics/operator/issues/2229 related to this. With the newest update the target is now VMAgent, but still needing .spec.shardCount set.