GitHub user hezeclark added a comment to the discussion: Override default_args
definition using passed params
Overriding `default_args` using passed params in Airflow requires understanding
when params are resolved. The key is that `params` are available at parse time
via `{{ params.key }}` in templates, but Python-level `default_args` are set at
DAG definition time.\n\n**Pattern 1: Use Jinja templating in operator
fields**\n```python\nfrom airflow import DAG\nfrom airflow.operators.python
import PythonOperator\nfrom datetime import datetime\n\nwith DAG(\n
dag_id='my_dag',\n params={'retries': 3, 'email': '[email protected]'},\n
default_args={'owner': 'airflow'},\n) as dag:\n # Use {{ params }} in
templates, not in default_args directly\n task = PythonOperator(\n
task_id='my_task',\n python_callable=my_function,\n retries='{{
params.retries }}', # templated\n )\n```\n\n**Pattern 2: Params in
`default_args` via a callable (Airflow 2.2+)**\n```python\n# Use Param objects
for structured params with defaults\nfrom airflow.models.param import
Param\n\nwit
h DAG(\n dag_id='my_dag',\n params={\n 'retry_count': Param(3,
type='integer', minimum=0, maximum=10),\n 'notify_email':
Param('[email protected]', type='string'),\n }\n) as dag:\n
pass\n```\n\n**Pattern 3: Dynamic default_args using the DAG run
conf**\n```python\n# In a PythonCallable, access conf directly\ndef
my_task(**context):\n params = context['params']\n retries =
params.get('retries', 3)\n # use runtime value\n```\n\nThe limitation is
that `default_args` like `retries` and `email_on_failure` are evaluated at
parse time, not run time. For runtime-configurable behavior, you need to access
`context['params']` inside the callable.
GitHub link:
https://github.com/apache/airflow/discussions/21627#discussioncomment-16154007
----
This is an automatically sent email for [email protected].
To unsubscribe, please send an email to: [email protected]
