Appearance
cluster.provision-gke ​
Provision a GKE cluster, register Kubernetes access, and optionally verify it.
Overview ​
| Property | Value |
|---|---|
| Workflow type | Dag |
| Library | App-clusters-gcp |
| Version | 1.0 |
Input Schema ​
| Field | Type | Required | Default | Description |
|---|---|---|---|---|
organization_uuid | uuid | Yes | — | Organization that will own the generated Kubernetes connection. |
gcp_connection_uuid | uuid | Yes | — | GCP CloudConnection UUID used for provider API calls. |
project_id | string | Yes | — | GCP project ID. |
location | string | Yes | — | GKE region or zone. |
cluster_name | string | Yes | — | GKE cluster name. |
display_name | string | No | — | Display name for the registered Kubernetes connection. |
network_name | string | No | — | VPC network name to validate and pass to GKE. |
subnetwork_name | string | No | — | Optional subnetwork name. |
kubernetes_version | string | No | — | Desired control-plane version. |
release_channel | string | No | — | GKE release channel such as REGULAR. |
initial_node_count | integer | No | — | Default node count for the legacy default pool when node_pools is omitted. |
machine_type | string | No | — | Default node machine type for the legacy pool or explicit node pools that omit machine_type. |
disk_size_gb | integer | No | — | Default node boot disk size for explicit node pools. |
node_service_account | string | No | — | Default service account email for nodes. |
node_pools | json | No | — | Explicit GKE node pools to create after control-plane readiness. Each item may declare name, node_count, machine_type, disk_size_gb, disk_type, image_type, service_account, oauth_scopes, labels, tags, taints, metadata, autoscaling, management, upgrade_settings, version, max_pods_per_node, spot, or preemptible. |
remove_default_node_pool | boolean | No | — | Remove the default GKE pool during cluster creation; defaults true when node_pools is supplied. |
labels | json | No | — | Labels for provider resources and connection metadata. |
private_cluster | boolean | No | — | Whether to create a private GKE cluster. |
deletion_protection | boolean | No | — | GKE deletion protection setting. |
register_connection | boolean | No | — | Register a Kubernetes CloudConnection after provisioning. |
verify_connection | boolean | No | — | Verify the registered Kubernetes CloudConnection. |
actor | string | No | — | Caller initiating the workflow. |
timeout_seconds | integer | No | — | Provider operation wait timeout. |
poll_interval | integer | No | — | Provider operation poll interval. |
Output Schema ​
| Field | Type | Required | Default | Description |
|---|---|---|---|---|
organization_uuid | uuid | No | — | Echoed organization UUID. |
gcp_connection_uuid | uuid | No | — | GCP CloudConnection UUID used. |
kubernetes_connection_uuid | uuid | No | — | Registered Kubernetes CloudConnection UUID. |
project_id | string | No | — | GCP project ID. |
location | string | No | — | GKE region or zone. |
cluster_name | string | No | — | GKE cluster name. |
display_name | string | No | — | Display name for the registered Kubernetes connection. |
network_name | string | No | — | VPC network name to validate and pass to GKE. |
subnetwork_name | string | No | — | Optional subnetwork name. |
kubernetes_version | string | No | — | Desired control-plane version. |
release_channel | string | No | — | GKE release channel such as REGULAR. |
initial_node_count | integer | No | — | Default node count for the legacy default pool when node_pools is omitted. |
machine_type | string | No | — | Default node machine type for the legacy pool or explicit node pools that omit machine_type. |
disk_size_gb | integer | No | — | Default node boot disk size for explicit node pools. |
node_service_account | string | No | — | Default service account email for nodes. |
node_pools | json | No | — | Explicit GKE node pools to create after control-plane readiness. |
remove_default_node_pool | boolean | No | — | Remove the default GKE pool during cluster creation. |
labels | json | No | — | Labels for provider resources and connection metadata. |
private_cluster | boolean | No | — | Whether to create a private GKE cluster. |
deletion_protection | boolean | No | — | GKE deletion protection setting. |
actor | string | No | — | Caller initiating the workflow. |
timeout_seconds | integer | No | — | Provider operation wait timeout. |
poll_interval | integer | No | — | Provider operation poll interval. |
workflow_run_uuid | string | No | — | Engine DAG run UUID for child step state. |
retried_from | string | No | — | Previous terminal state restored by engine retry. |
retry_count | integer | No | — | Number of retry attempts recorded by the engine. |
endpoint | string | No | — | Kubernetes API endpoint. |
provider_resource_id | string | No | — | Provider-native GKE cluster resource ID. |
provider_cluster_ref | string | No | — | Provider-neutral alias for the GKE cluster resource ID. |
status | string | No | — | Final provider cluster status. |
project_describe | json | No | — | Output from gcp.compute.project.describe. |
network_describe | json | No | — | Output from gcp.compute.network.describe. |
create_cluster | json | No | — | Output from gcp.gke.cluster.create. |
wait_cluster | json | No | — | Output from gcp.gke.operation.wait. |
create_node_pools | json | No | — | Batch output from gcp.gke.node_pools.create. |
wait_node_pools | json | No | — | Batch output from gcp.gke.node_pools.wait_ready. |
build_kubeconfig | json | No | — | Output from gcp.gke.build_kubeconfig; carries cluster metadata (api_server, ca_certificate) only. |
register_connection | json | No | — | Output from cluster.register-kubernetes-connection. |
verify_connection | json | No | — | Output from cluster.verify-kubernetes-connection. |
compensation_trigger | json | No | — | DAG compensation trigger. |
comp_current_layer | json | No | — | Current compensation layer. |
comp_queue | json | No | — | Remaining compensation queue. |
comp_retry_count | json | No | — | Current compensation retry count. |
failure_reason | string | No | — | Engine failure reason. |
failed_step | json | No | — | Failed DAG step. |
failed_layer | json | No | — | Failed DAG layer. |
failed_at_state | json | No | — | Failed state metadata. |
error | string | No | — | Error message. |
error_type | string | No | — | Error class. |
DAG Layers ​
| # | Layer | Steps | Compensation |
|---|---|---|---|
| 1 | preflight | gcp.compute.project.describe, gcp.compute.network.describe | — |
| 2 | create_cluster | gcp.gke.cluster.create | cluster.delete-gke |
| 3 | wait_cluster | gcp.gke.operation.wait | — |
| 4 | create_node_pools | gcp.gke.node_pools.create | cluster.delete-gke |
| 5 | wait_node_pools | gcp.gke.node_pools.wait_ready | — |
| 6 | connection_material | gcp.gke.build_kubeconfig | — |
| 7 | register | cluster.register-kubernetes-connection | cluster.delete |
| 8 | verify | cluster.verify-kubernetes-connection | — |
Execution Flow ​
Compensation ​
When any layer fails, its compensation steps run in reverse order to roll back the work completed so far.
Sub-workflows ​
| Sub-workflow | Step name |
|---|---|
gcp.compute.project.describe | project_describe |
gcp.compute.network.describe | network_describe |
gcp.gke.cluster.create | create_cluster |
gcp.gke.operation.wait | wait_cluster |
gcp.gke.node_pools.create | create_node_pools |
gcp.gke.node_pools.wait_ready | wait_node_pools |
gcp.gke.build_kubeconfig | build_kubeconfig |
cluster.register-kubernetes-connection | register_connection |
cluster.verify-kubernetes-connection | verify_connection |
cluster.delete-gke | delete_gke_cluster |
cluster.delete | delete_registered_cluster |
API Usage ​
bash
POST /api/workflows/start
Content-Type: application/json
{
"workflow_type": "cluster.provision-gke",
"initial_data": {
"organization_uuid": "value",
"gcp_connection_uuid": "value",
"project_id": "value",
"location": "value"
}
}