Azure ScaleCluster

Azure ScaleCluster

Certified

Update the autoscale settings of an Azure Machine Learning compute cluster

Sets the min/max node count (and optional idle scale-down delay) of an existing AmlCompute cluster. Only compute clusters support autoscaling; compute instances do not.

yaml
type: io.kestra.plugin.azure.ml.ScaleCluster
yaml
id: azure_ml_scale_cluster
namespace: company.team

tasks:
  - id: scale_up
    type: io.kestra.plugin.azure.ml.ScaleCluster
    tenantId: "{{ secret('AZURE_TENANT_ID') }}"
    clientId: "{{ secret('AZURE_CLIENT_ID') }}"
    clientSecret: "{{ secret('AZURE_CLIENT_SECRET') }}"
    subscriptionId: "{{ secret('AZURE_SUBSCRIPTION_ID') }}"
    resourceGroupName: ml-rg
    workspaceName: ml-workspace
    computeName: gpu-cluster
    minNodeCount: 1
    maxNodeCount: 4
Properties

Compute cluster name

Maximum node count

Maximum number of nodes the cluster can scale out to

Resource group name

Resource group containing the Machine Learning workspace

Subscription ID

Azure subscription GUID that owns the Machine Learning workspace

Azure AD tenant ID (GUID)

Workspace name

Name of the Azure Machine Learning workspace

Client ID of the Azure AD application

Application (client) ID used for service principal authentication.

Client secret for the Azure AD application

Secret value associated with the client ID; store in a Kestra secret.

Default0

Minimum node count

Minimum number of nodes kept available; 0 (default) scales the cluster down to no nodes when idle

Node idle time before scale down

How long a node stays idle before being deallocated; defaults to Azure's own setting when not provided

PEM-encoded certificate content for client authentication

PEM text for certificate-based auth; alternative to clientSecret.

Compute cluster name

Maximum node count

Minimum node count