Tableau Cloud: Airflow → Orchestra Conversion

Overview

Airflow's TableauRefreshWorkbookOperator (or TableauOperator) triggers a Tableau workbook or extract refresh and optionally waits for it to complete. TableauJobStatusSensor polls a running job. In Orchestra the equivalent is a task under the TABLEAU_CLOUD integration.

Parameter Mapping

Orchestra YAML Structure

version: v1
name: <pipeline-name>
pipeline:
  <stage-uuid>:
    tasks:
      <task-uuid>:
        integration: TABLEAU_CLOUD
        integration_job: TABLEAU_REFRESH_WORKBOOK
        name: <task_id value from Airflow>
        connection: <orchestra-tableau-cloud-connection-name>
        parameters:
          project_name: <tableau-project-name>   # required
          workbook_name: <tableau-workbook-name>    # required
        depends_on: []
        condition: null
        tags: []

Datasource refresh — if refreshing a published datasource instead of a workbook, use integration_job: TABLEAU_REFRESH_EXTRACT and parameters.datasource_name.

Conversion Steps

1. Identify the Airflow task — locate TableauRefreshWorkbookOperator or TableauOperator. Note workbook_name, site_id, and tableau_conn_id.
2. Create/verify the Orchestra connection — in Orchestra Settings → Connections, create a Tableau Cloud connection with server URL, site name, and either a Personal Access Token (PAT) or a Connected App (direct trust JWT: client ID, secret ID, secret value, and the Tableau username to sign in as). The site is configured on the connection, not per-task.
3. Replace operator with task block — use the YAML above.
4. Drop any TableauJobStatusSensor — Orchestra's task already polls for completion.
5. Wire dependencies — convert >> chains to depends_on:.

Before / After Example

Airflow DAG (before)

from airflow.providers.tableau.operators.tableau import TableauOperator

refresh_sales_wb = TableauOperator(
    task_id="refresh_sales_dashboard",
    resource="workbooks",
    method="refresh",
    find="Sales Dashboard",
    match_with="name",
    tableau_conn_id="tableau_cloud_prod",
    blocking_refresh=True,
)

Orchestra YAML (after)

version: v1
name: my-pipeline
pipeline:
  stage-001:
    tasks:
      task-001:
        integration: TABLEAU_CLOUD
        integration_job: TABLEAU_REFRESH_WORKBOOK
        name: refresh_sales_dashboard
        connection: tableau_cloud_prod_12345
        parameters:
          project_name: Sales           # required — Tableau project containing the workbook
          workbook_name: Sales Dashboard
        depends_on: []
        condition: null
        tags: []

Gotchas

• Workbook name vs ID: Orchestra uses the workbook display name. Ensure it matches exactly (case-sensitive) as it appears in Tableau Cloud.
• Site ID on the connection: unlike Airflow where site_id is a task parameter, Orchestra stores it on the connection. If you refresh workbooks across multiple Tableau sites, you need one Orchestra connection per site.
• Auth: PAT or Connected App: the Orchestra Tableau connection takes tableau_auth_type PAT (token name + value) or CONNECTED_APP (client ID, secret ID, secret value, username). A PAT allows only one in-flight request at a time, so prefer a Connected App when several tasks refresh concurrently through one connection. Connected Apps need Tableau Cloud (Dec 2025 release+) or Tableau Server 2025.3+. Username/password auth isn't an option — move to one of these when creating the Orchestra connection.
• TableauOperator with resource="datasources": map to TABLEAU_REFRESH_EXTRACT in Orchestra, using parameters.datasource_name instead of workbook_name.
• blocking_refresh=False + sensor pattern: both collapse into one Orchestra task.
• Tableau Server (on-prem): this skill is for Tableau Cloud. For Tableau Server, confirm whether Orchestra's Tableau Cloud integration supports your Server version via the REST API.

References

• Orchestra docs: https://docs.getorchestra.io/docs/integrations/tableau_cloud
• Airflow Tableau provider: https://airflow.apache.org/docs/apache-airflow-providers-tableau/stable/operators/tableau.html

Adding Alerts

If the Airflow DAG uses on_failure_callback or on_success_callback for Slack/email notifications, replace those with an alerts block in the pipeline YAML. Alerts fire based on overall pipeline status and support Slack, Email, PagerDuty, Microsoft Teams, and Webhook destinations.

version: v1
name: my-pipeline

alerts:
  - name: on-failure
    statuses:
      - FAILED
    destinations:
      - integration: SLACK
        destination: '#data-alerts'
    custom_message: 'Optional context message.'

  - name: on-success
    statuses:
      - SUCCEEDED
    destinations:
      - integration: SLACK
        destination: '#data-alerts'

pipeline:
  # ... tasks unchanged

Valid statuses: FAILED, SUCCEEDED, CANCELLED, WARNING, SKIPPED. Multiple alerts with different destinations are supported — each needs a unique name. See the slack-airflow-to-orchestra skill for full schema details.