Quotas

Every project has a limit per resource type. A request that would take the project over one of its limits is refused with 403 Forbidden and the quota_exceeded error; nothing is created. Deleting or stopping resources frees their share again.

Limits apply per project. To raise one, contact support with the project ID and the quota key you need more of.

Quota keys

Key	Counts	Unit
cpu-dedicated	vCPUs of instances	core
ram-dedicated	Memory of instances	GiB
gpu-<model>	GPUs of one model, e.g. gpu-nvidia-h100-sxm5	gpu
volumes-ssd	Size of SSD volumes	GiB
volumes-hdd	Size of HDD volumes	GiB
snapshots-ssd	Size of SSD snapshots	GiB
snapshots-hdd	Size of HDD snapshots	GiB
s3-storage	Data stored by the project's S3 users	byte
ipv4	Public IPv4 addresses	address
load-balancers	Load balancers	count
kubernetes-clusters	Kubernetes clusters	count

A snapshot counts with the provisioned size of its source volume at the time it was taken.

s3-storage is enforced by the object storage itself rather than by the API: it is applied as the storage quota of the project's S3 users, and uploads beyond it are rejected by the S3 endpoint.

Quota Schema

• key String - The quota key, see Quota keys.
• capacity Integer - The project's limit for this key. -1 means unlimited.
• used Integer - How much of the limit is currently in use.
• flavors Array - The flavors currently drawing on this key and how much each uses. Recomputed periodically, so it can briefly lag the key's used.
  • flavor_name String - The instance type for compute keys (e.g. vcpu-4_memory-8g), the storage class for storage keys, or the kind of consumer (ephemeral, floating, load-balancer, kubernetes) for the rest.
  • product_name String - Human-readable name of the resource, e.g. vCPU.
  • unit String - The unit of used, see Quota keys.
  • used Integer - How much this flavor uses.

used can be higher than capacity when a limit was lowered after the resources were created. Existing resources are kept; new requests for that key are refused until usage drops below the limit.

Get Quotas

Returns the active project's limit and current usage for every quota key.

GET /compute/v1/quotas (HTTP 200 - OK)

Response body

{
  "quotas": [
    {
      "key": "cpu-dedicated",
      "capacity": 4,
      "used": 4,
      "flavors": [
        {
          "flavor_name": "vcpu-4_memory-8g",
          "product_name": "vCPU",
          "unit": "core",
          "used": 4
        }
      ]
    },
    {
      "key": "volumes-ssd",
      "capacity": 240,
      "used": 0,
      "flavors": []
    }
  ]
}

Quota Exceeded Error

A create or resize request that would exceed a limit returns 403 Forbidden. code is always quota_exceeded, and quota says which limit was hit:

• key String - The quota key that would be exceeded, as listed by Get Quotas.
• limit Integer - The project's limit for this key.
• used Integer - How much of the limit is already in use.
• requested Integer - How much the rejected request asked for.

{
  "code": "quota_exceeded",
  "message": "Project quota exceeded for 'cpu-dedicated': requested 4, 4 of 4 already in use",
  "quota": {
    "key": "cpu-dedicated",
    "limit": 4,
    "used": 4,
    "requested": 4
  }
}

Match on code and quota.key rather than on message, whose text may change.