CapacityQuota

Preview

This feature is subject to the "Pre-GA Offerings Terms" in the General Service Terms section of the Service Specific Terms. Pre-GA features are available "as is" and might have limited support. For more information, see the launch stage descriptions.

The CapacityQuota CustomResourceDefinition (CRD) limits the amount of resources that can be provisioned for specific subsets of nodes that you target by using label selectors. For more information about how CapacityQuotas work, see Configure granular resource limits.

apiVersion: autoscaling.x-k8s.io/v1beta1
kind: CapacityQuota
metadata:
  name: capacity-quota
spec:
  selector:
    matchLabels:
      cloud.google.com/compute-class: my-class
    matchExpressions:
      - key: cloud.google.com/machine-family
        operator: In
        values:
          - c2
          - c3
          - c3d
  limits:
    resources:
      cpu: 64
      memory: 128Gi
status:
  observedGeneration: 1
  conditions:
  - lastTransitionTime: "2026-07-27T10:00:00Z"
    message: "CapacityQuota is valid"
    reason: "Valid"
    status: "True"
    type: "cluster-autoscaler.kubernetes.io/valid"
  used:
    resources:
      cpu: "32"
      memory: 128Gi

CapacityQuota specification

spec:
  limits: object (limits)
  selector: object (selector)

Fields
spec.limits required	object (limits) This object defines resource limits for the matched nodes.
spec.selector optional	object (selector) Label selectors to target subsets of nodes. If you omit this field or you don't specify a value, then the CapacityQuota matches all of the nodes in the cluster.

limits

limits object defines the resource limits of a CapacityQuota.

limits:
  resources:
    string: string

Fields

resources required

map (string: string) The `resources` map define the resource limits of this quota. Map keys are resource names and map values are strings in the resource.Quantity format. The following resource names are supported: cpu: limits for CPU resources. memory: limits for memory resources. nodes: limits for the number of nodes. nvidia.com/gpu: limits for the number of NVIDIA GPUs. google.com/tpu: limits for the number of TPUs. All resource quantities must be integer values that are greater than 0. You can specify binary and decimal values if those values can be converted to integers. Milli units (m) aren't supported, even if the milli unit represents an integer value. Examples of valid resource quantities include "32Gi", "2k", and "10.0". Examples of invalid quantities include "1000m", "3.67Gi", and "0.3".

selector

This is a standard Kubernetes LabelSelector object.

matchExpressions:
  - object (matchExpressions)
matchLabels:
  string: string

Fields
matchExpressions[] optional	object (matchExpressions) A list of label selector requirements. The requirements are ANDed.
matchLabels optional	map (string: string) A map of key-value pairs. A single entry in the `matchLabels` map is equivalent to an element of `matchExpressions` field, whose key field is "key", the operator is "In", and the values array contains only "value". The requirements are ANDed.

matchExpressions

This is a standard Kubernetes LabelSelectorRequirement object.

key: string
operator: string
values:
  - string

Fields
key required	string The label key that the selector applies to.
operator required	string This represents a key's relationship to a set of values. Valid operators are In, NotIn, Exists and DoesNotExist.
values optional	array (string) This is an array of string values. If the operator is In or NotIn, the values array must be non-empty. If the operator is Exists or DoesNotExist, the values array must be empty. This array is replaced during a strategic merge patch.

status

The status object indicates the observed state of the CapacityQuota resource.

conditions:
  - object (Condition)
observedGeneration: integer
used: object (used)

Fields
conditions[]	object (Condition) The `conditions` array provide a standard mechanism for reporting the quota's state. Cluster Autoscaler manages the cluster-autoscaler.kubernetes.io/valid condition, and will enforce the quota only if the status of the condition is True.
observedGeneration	integer The `observedGeneration` represents the last `.metadata.generation` observed by the controller.
used	object (used) The `used` object shows the current usage of the quota.

used

The used object shows the current usage of the quota.

resources:
  string: string

Fields
resources	map (string: string) The `resources` map shows the current usage of the resources defined in the quota limits.