Resize or Autoscale Node Pool

PUT

/tenant_clusters/{tenant_cluster_id}/node-pools/{node_pool_id}/autoscaling/

curl --request PUT \
  --url 'https://api.e2enetworks.com/myaccount/api/v1/gpu/tenant_clusters/{tenant_cluster_id}/node-pools/{node_pool_id}/autoscaling/?project_id=<PROJECT_ID>&location=<LOCATION>&apikey=<API_KEY>' \
  --header 'Authorization: Bearer <API_TOKEN>' \
  --header 'Content-Type: application/json' \
  --data '{
  "min_replicas": 0,
  "max_replicas": 0,
  "triggers": [
    {
      "type": "cpu",
      "target_utilization_percent": 0
    }
  ],
  "polling_interval_seconds": 0,
  "scale_up_stabilization_window_seconds": 0,
  "scale_down_stabilization_window_seconds": 0
}'

Sets a worker pool’s size, in place of whatever it had. Send equal min_replicas and max_replicas and no triggers for a fixed size: the pool shows resizing until it reaches that count. Send a lower minimum than maximum, with a CPU target, a memory target, or both, to autoscale: the pool runs as many nodes as its busiest target asks for, within the bounds, and shows scaling while its autoscaler changes the count. The cluster stays ready throughout.

The pool must be ready and on an hourly plan; a committed pool cannot be resized or autoscaled. Growing a pool checks credit and free stock for the added nodes. Scaling down removes nodes and evicts their pods.

Parameters

Path Parameters

tenant_cluster_id

required

integer

Kubernetes cluster ID, as returned by List Kubernetes Clusters.

node_pool_id

required

integer

Worker node pool ID, from the cluster’s node_pools list.

Query Parameters

project_id

required

integer

Project ID

active_iam

integer

Active IAM ID (To access contact person account) Find your Active IAM ID here

location

required

string

Allowed values: Delhi Chennai

Location

Request Body^required

Media typeapplication/json

object

min_replicas

required

The fewest nodes. Equal to max_replicas for a fixed size.

integer

>= 1 <= 25

max_replicas

required

The most nodes.

integer

>= 1 <= 25

triggers

Required when autoscaling, and empty or left out for a fixed size. At most one cpu and one memory target.

Array<object>

<= 2 items

object

type

required

string

Allowed values: cpu memory

target_utilization_percent

The average use across the pool’s nodes to aim for. Defaults to 70 for cpu and 80 for memory.

integer

>= 20 <= 90

polling_interval_seconds

How often the targets are checked, when autoscaling. null keeps the default, 30.

integer

nullable >= 15 <= 300

scale_up_stabilization_window_seconds

How long use must stay above the target before nodes are added, when autoscaling. null keeps the default, 0.

integer

nullable <= 3600

scale_down_stabilization_window_seconds

How long use must stay below the target before nodes are removed, when autoscaling. null keeps the default, 300.

integer

nullable <= 3600

Responses

200

The new size is saved.

Media typeapplication/json

object

code

integer

data

The pool’s size settings, as in Get Node Pool Size, without available_trigger_types.

object

node_pool_id

integer

min_replicas

integer

max_replicas

integer

replicas

integer

status

string

errors

object

message

string

Example

{
  "code": 200,
  "data": {
    "node_pool_id": 22,
    "min_replicas": 4,
    "max_replicas": 4,
    "replicas": 4,
    "status": "resizing"
  },
  "errors": {},
  "message": "Node pool resize to 4 nodes queued successfully"
}