Params¶

Params enable you to provide runtime configuration to tasks. You can configure default Params in your DAG code and supply additional Params, or overwrite Param values, at runtime when you trigger a DAG. Param values are validated with JSON Schema. For scheduled DAG runs, default Param values are used.

DAG-level Params¶

To add Params to a DAG, initialize it with the params kwarg. Use a dictionary that maps Param names to either a Param or an object indicating the parameter's default value.

 from airflow import DAG
 from airflow.models.param import Param

 with DAG(
     "the_dag",
     params={
         "x": Param(5, type="integer", minimum=3),
         "y": 6
     },
 ):

Task-level Params¶

You can also add Params to individual tasks.

PythonOperator(
    task_id="print_x",
    params={"x": 10},
    python_callable=print_it,
)

Task-level params take precedence over DAG-level params, and user-supplied params (when triggering the DAG) take precedence over task-level params.

Referencing Params in a Task¶

Params can be referenced in templated strings under params. For example:

 PythonOperator(
     task_id="from_template",
     op_args=[
         "{{ params.int_param + 10 }}",
     ],
     python_callable=(
         lambda x: print(x)
     ),
 )

Even though Params can use a variety of types, the default behavior of templates is to provide your task with a string. You can change this by setting render_template_as_native_obj=True while initializing the DAG.

 with DAG(
     "the_dag",
     params={"x": Param(5, type="integer", minimum=3)},
     render_template_as_native_obj=True
 ):

This way, the Param's type is respected when it's provided to your task:

# prints <class 'str'> by default
# prints <class 'int'> if render_template_as_native_obj=True
PythonOperator(
    task_id="template_type",
    op_args=[
        "{{ params.int_param }}",
    ],
    python_callable=(
        lambda x: print(type(x))
    ),
)

Another way to access your param is via a task's context kwarg.

 def print_x(**context):
     print(context["params"]["x"])

 PythonOperator(
     task_id="print_x",
     python_callable=print_x,
 )

JSON Schema Validation¶

Param makes use of JSON Schema, so you can use the full JSON Schema specifications mentioned at https://json-schema.org/draft/2020-12/json-schema-validation.html to define Param objects.

with DAG(
    "my_dag",
    params={
        # an int with a default value
        "int_param": Param(10, type="integer", minimum=0, maximum=20),

        # a required param which can be of multiple types
        # a param must have a default value
        "dummy": Param(5, type=["null", "number", "string"]),

        # an enum param, must be one of three values
        "enum_param": Param("foo", enum=["foo", "bar", 42]),

        # a param which uses json-schema formatting
        "email": Param(
            default="example@example.com",
            type="string",
            format="idn-email",
            minLength=5,
            maxLength=255,
        ),
    },
):

Note

As of now, for security reasons, one can not use Param objects derived out of custom classes. We are planning to have a registration system for custom Param classes, just like we've for Operator ExtraLinks.

Disabling Runtime Param Modification¶

The ability to update params while triggering a DAG depends on the flag core.dag_run_conf_overrides_params. Setting this config to False will effectively turn your default params into constants.