> ## Documentation Index > Fetch the complete documentation index at: https://docs.ornn.com/llms.txt > Use this file to discover all available pages before exploring further. # Create a cluster > Launches a Kubernetes or Slurm cluster on capacity your organization already holds; it does not purchase GPUs. List the hosts to use in node_ids. Each must be in bare-metal status, not in another cluster, and on an active reservation your organization owns; hosts can come from more than one such reservation. If node_ids and virtual_machine_ids are both omitted or empty, every bare-metal host on reservation_id is used. Before a host joins, Ornn prepares it - running processes are stopped, the tenant home directory is wiped, and any VM on that host is removed - so save its data first. Your organization can run one cluster of each kind at a time, and cannot run Kubernetes and Slurm clusters at the same time. A cluster counts until its teardown finishes. A repeat request for a reservation whose cluster is active, or provisioning with workers attached, returns that cluster with 202 and changes nothing, even if the node list differs; change membership with the worker operations. A request after a failed launch on the same reservation relaunches it under the same id. Do not create while a teardown is in progress. A 202 response means the launch was accepted, not that the cluster is ready. Poll listClusters until state is active or failed, then call getClusterAccess. This route does not read an Idempotency-Key header; after a timeout or 5xx, list clusters and look for the reservation before sending the request again. Unknown body fields are rejected. Malformed JSON, unknown fields, or a missing Content-Type return 400, 422, or 415 with a plain-text body. For the full lifecycle with polling and retry guidance, follow [Automate managed Kubernetes with the Ornn API](/guides/kubernetes-api) or [Launch and manage Slurm through the API](/guides/slurm-access#launch-and-manage-through-the-api). CLI equivalent, after `ornn login`: ```bash theme={null} ornn launch --mode kubernetes --reservation RESERVATION_ID --nodes 1 \ --machines NODE_ID --network public ``` ## OpenAPI ````yaml openapi.yaml POST /v1/orchestrator/controllers openapi: 3.1.0 info: title: Ornn compute API version: 1.0.0 description: >- Customer HTTP API for Ornn GPU capacity. It covers reserved capacity and nodes, VM and Bare Metal access, Kubernetes and Slurm clusters, SSH keys, storage volumes and GitHub repositories, private networks, monitoring, buying and selling capacity (Reserve, Market, Spot, and Relist), and billing. API keys are scoped to one organization and act with their creator's access there; account, team, and API-key management stay in the console. Most operations require an approved organization. See https://docs.ornn.com/api-keys to create a key and https://docs.ornn.com/guides/kubernetes-api for an end-to-end Kubernetes integration. Examples use fictional IDs and credentials. The info.version field identifies the contract revision, not a shared API path version. contact: email: support@ornn.com servers: - url: https://api.ornn.com description: Production security: - OrnnApiKey: [] tags: - name: Authentication - name: Reservations - name: GPU nodes - name: Clusters - name: Workers - name: Virtual machines description: >- Choose a VM disk, launch VM access on a reservation, and read each VM's status, launch progress, and SSH connection details. - name: Bare Metal description: >- Launch Bare Metal access on a reservation, read a host's readiness and SSH connection details, and reboot or hard reset it. - name: SSH keys description: >- Register public keys for your organization, choose which keys a VM or Bare Metal reservation uses, and queue key delivery to machines. Slurm clusters take organization key IDs at launch. Kubernetes API access does not use SSH keys. - name: Reserve description: >- Browse reserved-capacity listings and their terms, and post, cancel, or accept offers. Reserve listings are visible to approved organizations. - name: Market description: >- Browse relisted capacity and buy windows at the asking price, paid up front. Open to every member of your organization. - name: Spot description: >- Hourly order books for VMs with 1, 2, 4, or 8 GPUs on one node: read books, trades, and quotes, place and cancel bids, and follow fills. - name: Relist description: >- Sell whole nodes from a reservation you hold at your asking price, and manage those listings. - name: Storage description: >- Managed volumes, their files, and GitHub repository pulls. Volumes belong to the API key's creator in the key's organization. - name: Networks description: >- Organization private networks and the WireGuard configs you use to reach private machines and clusters. - name: Monitoring description: >- Machine telemetry and console logs for capacity your organization holds. Reads are limited to your reservation window. - name: Billing description: >- Invoices, billing summary, usage cost, Ornn credits, and account statements. Saving a card and paying invoices stay in the console. externalDocs: description: REST integration walkthrough url: >- https://docs.ornn.com/guides/kubernetes-access#provision-and-manage-through-rest paths: /v1/orchestrator/controllers: post: tags: - Clusters summary: Create a cluster description: >- Launches a Kubernetes or Slurm cluster on capacity your organization already holds; it does not purchase GPUs. List the hosts to use in node_ids. Each must be in bare-metal status, not in another cluster, and on an active reservation your organization owns; hosts can come from more than one such reservation. If node_ids and virtual_machine_ids are both omitted or empty, every bare-metal host on reservation_id is used. Before a host joins, Ornn prepares it - running processes are stopped, the tenant home directory is wiped, and any VM on that host is removed - so save its data first. Your organization can run one cluster of each kind at a time, and cannot run Kubernetes and Slurm clusters at the same time. A cluster counts until its teardown finishes. A repeat request for a reservation whose cluster is active, or provisioning with workers attached, returns that cluster with 202 and changes nothing, even if the node list differs; change membership with the worker operations. A request after a failed launch on the same reservation relaunches it under the same id. Do not create while a teardown is in progress. A 202 response means the launch was accepted, not that the cluster is ready. Poll listClusters until state is active or failed, then call getClusterAccess. This route does not read an Idempotency-Key header; after a timeout or 5xx, list clusters and look for the reservation before sending the request again. Unknown body fields are rejected. Malformed JSON, unknown fields, or a missing Content-Type return 400, 422, or 415 with a plain-text body. operationId: createCluster requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/CreateCluster' examples: kubernetes: summary: Kubernetes on one selected host value: kind: kubernetes reservation_id: 11111111-1111-4111-8111-111111111111 network_mode: public node_ids: - 33333333-3333-4333-8333-333333333333 slurm: summary: Slurm on two selected hosts with one SSH key value: kind: slurm reservation_id: 11111111-1111-4111-8111-111111111111 network_mode: public node_ids: - 33333333-3333-4333-8333-333333333333 - 77777777-7777-4777-8777-777777777777 ssh_key_ids: - 88888888-8888-4888-8888-888888888888 responses: '202': description: >- Launch accepted. node_count and gpu_count count workers attached so far and are usually 0 in this response. Poll listClusters for progress. content: application/json: schema: $ref: '#/components/schemas/ClusterAccepted' example: id: 44444444-4444-4444-8444-444444444444 kind: kubernetes organization_id: 22222222-2222-4222-8222-222222222222 access_mode: kubernetes state: provisioning network_mode: public endpoint: null credential_state: pending namespace: k8s-22222222-2222-4222-8222-222222222222 node_count: 0 gpu_count: 0 progress: null created_at: '2026-10-14T12:00:00Z' updated_at: '2026-10-14T12:00:00Z' '400': description: The request names an unsupported cluster kind or repeats a host. content: application/json: schema: $ref: '#/components/schemas/Error' examples: unsupportedKind: summary: Unsupported cluster kind value: detail: code: unsupported_controller_kind message: >- That cluster type is not supported. Choose Kubernetes or Slurm. duplicateNode: summary: The same host is listed twice value: detail: code: cluster_nodes_duplicate message: >- Each node can only be selected once. Review the node selection and try again. '401': $ref: '#/components/responses/Unauthorized' '403': $ref: '#/components/responses/Forbidden' '404': description: >- The reservation, a listed host, or a listed VM is not visible to your organization. A supplier's view of a sale does not grant launch rights. content: application/json: schema: $ref: '#/components/schemas/Error' examples: reservation: summary: Reservation not owned or not found value: detail: code: reservation_not_found message: Ornn could not find that reservation. node: summary: A listed host is not owned or does not exist value: detail: code: node_not_found message: Ornn could not find that node in this cluster. virtualMachine: summary: A listed VM is not owned, does not exist, or is offline value: detail: code: virtual_machine_not_found message: Ornn could not find that virtual machine. '409': description: >- A lifecycle or capacity conflict prevents launch. Read detail.code and current state before retrying. launch_node_offline names the host and its last heartbeat at the top level. cluster_access_mode_conflict adds live_mode, the kind of cluster your organization already runs. content: application/json: schema: $ref: '#/components/schemas/Error' examples: otherKindRunning: summary: A cluster of the other kind is still live value: detail: code: cluster_access_mode_conflict message: >- This reservation already has a live cluster in another access mode. Tear it down before launching a different cluster type. live_mode: slurm otherClusterRunning: summary: >- Your organization already runs a Kubernetes cluster on another reservation value: detail: code: kubernetes_cluster_ambiguous_group message: >- These nodes belong to more than one Kubernetes cluster. Choose nodes from one cluster. nodeOffline: summary: A selected host has not sent a heartbeat recently value: detail: code: launch_node_offline message: >- That node is offline. Power it on and try the launch again. node_id: 33333333-3333-4333-8333-333333333333 last_heartbeat_at: '2026-10-14T11:40:00+00:00' heartbeat_max_age_seconds: 180 nodesUnavailable: summary: A selected host is not in bare-metal status value: detail: code: cluster_nodes_not_available message: >- One or more selected nodes are no longer available. Refresh the node list and try again. reservationEnded: summary: >- The reservation ended, was cancelled, or has a pending change value: detail: code: reservation_not_launchable message: This reservation cannot launch in its current state. sameHost: summary: A host and a VM on that host were both listed value: detail: code: node_virtual_machine_same_host message: Choose that node or its virtual machine, not both. slurmKeyRequired: summary: Slurm launch with no active organization SSH key value: detail: code: slurm_ssh_key_required message: >- Add an SSH key to your account first. Slurm logins use your saved account SSH keys. slurmNodeCount: summary: Slurm node_count does not match the hosts that resolved value: error: cluster_nodes_requested_unavailable detail: cluster_nodes_requested_unavailable requested: 2 available: 1 '422': description: >- Invalid cluster configuration. The example uses an unsupported network mode. content: application/json: schema: $ref: '#/components/schemas/Error' example: detail: code: invalid_network_mode message: Choose Public or Private VPN for the cluster network. default: $ref: '#/components/responses/Error' components: schemas: CreateCluster: type: object required: - kind - reservation_id properties: kind: type: string enum: - kubernetes - slurm description: The cluster type to launch. reservation_id: type: string format: uuid description: >- An active reservation your organization holds. A repeat request for the same reservation returns the same cluster. Use reservationId from the reservation list, not a listing ID. network_mode: type: string enum: - public - private default: public description: >- public exposes the Kubernetes API or Slurm SSH login on the internet, with authentication. private requires a WireGuard connection to your organization's private network. network_id: type: - string - 'null' format: uuid description: >- An existing organization network for private access. Used only when network_mode is private. Omit it to use the shared default network; this does not create a network. node_ids: type: array items: type: string format: uuid description: >- Host IDs (node_id from the reservation node list), not GPU IDs. Each host must be in bare-metal status, not in another cluster, and on an active reservation your organization owns. Ornn prepares each host before it joins, which stops its processes, wipes the tenant home directory, and removes any VM on it. If this and virtual_machine_ids are both omitted or empty, every bare-metal host on reservation_id is used. virtual_machine_ids: type: array items: type: string format: uuid description: >- Existing VMs on reservations your organization owns to use as workers. A VM must not be deleted, destroyed, or offline. Do not list a VM and its host in node_ids in the same request. node_count: type: - integer - 'null' description: >- Optional expected host count; it never selects or limits hosts. Kubernetes ignores it. Slurm rejects the launch with 409 cluster_nodes_requested_unavailable when it differs from the number of hosts that resolved. Omit it, or send exactly the number of hosts you list. ssh_key_ids: type: array items: type: string format: uuid description: >- Slurm only. Organization SSH key IDs (ssh_keys[].id from GET /v1/organizations/{organization_id}/ssh-keys) to authorize on the login node. Omit to authorize every active key in your organization; with no active key, Slurm launch returns 409 slurm_ssh_key_required. Kubernetes ignores this field. cluster_name: type: - string - 'null' deprecated: true description: Accepted for compatibility and ignored. additionalProperties: false ClusterAccepted: type: object required: - id - state properties: id: type: string format: uuid kind: type: string organization_id: type: - string - 'null' format: uuid access_mode: type: string state: type: string network_mode: type: string endpoint: type: - string - 'null' credential_state: type: string namespace: type: - string - 'null' description: The namespace Ornn configures for your workloads. node_count: type: integer description: >- Workers attached when this response was built; usually 0 right after a launch. gpu_count: type: integer description: GPUs on the workers counted in node_count. progress: anyOf: - $ref: '#/components/schemas/Progress' - type: 'null' created_at: type: string format: date-time updated_at: type: string format: date-time Error: type: object properties: code: type: string detail: oneOf: - type: string - type: object required: - code - message properties: code: type: string message: type: string Progress: type: object description: >- Stages of the latest launch or teardown. Launch stages are create_plane, prepare_hosts, join_workers, apply_plane, and issue_access. Teardown stages include workers and the control-plane stage (namespace for Kubernetes, stack for Slurm), plus nodes or storage when those steps fail. properties: stages: type: array items: type: object properties: key: type: string description: Stable stage identifier. label: type: string description: Display label for the stage. status: type: string enum: - pending - running - completed - failed detail: type: - string - 'null' description: Failure reason or worker confirmation count, when available. responses: Unauthorized: description: >- The credential is missing, invalid, revoked, or no longer belongs to a current organization member. Check the HTTP status before parsing the body. content: application/json: schema: $ref: '#/components/schemas/Error' Forbidden: description: >- The credential lacks access to this operation, or the organization requires approval. An API key does not bypass these checks. content: application/json: schema: $ref: '#/components/schemas/Error' Error: description: >- Unsuccessful request. Check HTTP status first. Errors may have detail.code/detail.message, a string detail and optional top-level code, or an empty/non-JSON body. 401: invalid credential; 403: access or approval required; 404: not visible or credentials not ready; 409: lifecycle conflict; 429: back off; 5xx: service failure. Inspect current state before retrying mutations. content: application/json: schema: $ref: '#/components/schemas/Error' securitySchemes: OrnnApiKey: type: http scheme: bearer bearerFormat: ornn_sk_... description: >- Create an organization API key in Account Settings > Developer. Do not send this key to the Kubernetes API server. ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.