Databricks Start Cluster Then Run Job

arazzo: 1.0.1
info:
  title: Databricks Start Cluster Then Run Job
  summary: Start a terminated cluster, wait until RUNNING, then trigger a job run.
  description: >-
    Starts a previously terminated Databricks cluster, polls until the cluster
    is RUNNING, and then triggers an immediate run of a job that runs on that
    cluster. The cluster_id drives the start and poll steps while the job_id
    drives the run-now request. Every step spells out its request inline so the
    flow can be read and executed without opening the underlying OpenAPI
    description.
  version: 1.0.0
  x-realizes-capability-ids:
  - BC-600.50
  x-capability-derivation:
    method: 'deterministic join: sourceDescriptions -> per-tag OpenAPI -> tag/capability edge. No classification at this step.'
    min_confidence: 0.7
    sources:
    - capability_id: BC-600.50
      capability_name: IT Infrastructure Management
      spec: databricks-clusters-api-openapi.yml
      confidence: 0.75
    model: Turbo EA Capabilities by Vincent Verdet — Turbo EA, https://github.com/vincentmakes/turbo-ea-capabilities, CC BY 4.0
sourceDescriptions:
- name: clustersApi
  url: ../openapi/databricks-clusters-api-openapi.yml
  type: openapi
- name: jobsApi
  url: ../openapi/databricks-jobs-api-openapi.yml
  type: openapi
workflows:
- workflowId: start-cluster-and-run-job
  summary: Start a terminated cluster, wait for RUNNING, then run a job.
  description: >-
    Starts the named cluster, polls its status until RUNNING, and then submits
    a run-now request for the supplied job.
  inputs:
    type: object
    required:
    - cluster_id
    - job_id
    properties:
      cluster_id:
        type: string
        description: The terminated cluster to start.
      job_id:
        type: integer
        description: The job to run once the cluster is RUNNING.
      notebook_params:
        type: object
        description: Optional notebook parameters for the job run.
  steps:
  - stepId: startCluster
    description: >-
      Start the terminated cluster using its prior configuration. Starting is
      asynchronous, so a poll loop confirms readiness.
    operationId: startCluster
    requestBody:
      contentType: application/json
      payload:
        cluster_id: $inputs.cluster_id
    successCriteria:
    - condition: $statusCode == 200
  - stepId: pollClusterState
    description: >-
      Read the cluster status and inspect the life cycle state. Loop back while
      the cluster is still PENDING; continue once it is RUNNING.
    operationId: getCluster
    parameters:
    - name: cluster_id
      in: query
      value: $inputs.cluster_id
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      state: $response.body#/state
    onSuccess:
    - name: stillPending
      type: goto
      stepId: pollClusterState
      criteria:
      - context: $response.body
        condition: $.state == "PENDING"
        type: jsonpath
    - name: running
      type: goto
      stepId: runJobNow
      criteria:
      - context: $response.body
        condition: $.state == "RUNNING"
        type: jsonpath
  - stepId: runJobNow
    description: >-
      Trigger the job now that the cluster is RUNNING. Returns the run_id of the
      triggered run.
    operationId: runJobNow
    requestBody:
      contentType: application/json
      payload:
        job_id: $inputs.job_id
        notebook_params: $inputs.notebook_params
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      runId: $response.body#/run_id
  outputs:
    clusterId: $inputs.cluster_id
    runId: $steps.runJobNow.outputs.runId