All Products
Search

This Product

    Data Management:Airflow DMS Operator

    Document Center

    Data Management:Airflow DMS Operator

    Last Updated:Jun 22, 2026

    DMS provides custom Airflow operators for secure access to DMS-managed resources. Choose the operators that suit your workflow.

    Prerequisites

    Procedure

    Note

    For more information, see Workflow Development.

    1. Navigate to the WORKSPACE or REPOS page to configure your code.

    2. Configure the code in your Python file.

      Operator

      Description

      DMSSqlOperator

      Submits SQL to a DMS-managed database instance and retrieves the results.

      DTSLakeInjectionOperator

      Uses DTS to sync data from a DMS-managed database to OSS.

      DMSNotebookOperator

      Runs a DMS-managed notebook file (.ipynb).

      DMSAnalyticDBSparkOperator

      Submits Spark SQL to a specific resource group managed by AnalyticDB for MySQL (Lakehouse Edition) (task type: Interactive; engine: Spark).

      DMSLockFreeSqlOperator

      Runs lock-free, large-scale DML operations through the DMS chunked DML service. Ideal for bulk UPDATE or DELETE operations on large tables.

      DMSLindormSparkOperator

      Runs SQL or JAR tasks on the Lindorm Spark engine.

    3. Run the Python file in the Airflow UI or using the REST API.

    Common features

    The following features apply to all DMS Airflow operators.

    Task cancellation

    All operators implement the on_kill() method. When Airflow sends a termination signal—for example, when a task is manually marked as failed or times out—the operator automatically calls the cancellation API of the corresponding cloud service to clean up remote tasks and prevent resource waste.

    Retry mechanism

    Task submission and status checks include built-in retries powered by tenacity:

    • Task submission: Up to 3 retries, with a 5-second interval.

    • Status check: Up to 5 retries, with a 3-second interval.