Default args and DAG parameters in Apache Airflow - Step-by-Step Execution

Process Flow - Default args and DAG parameters

Define default_args dict

↓

Create DAG with default_args

↓

Define tasks using DAG

↓

Run DAG

↓

Tasks inherit default_args

↓

Override parameters if needed

↓

DAG executes with parameters

This flow shows how default arguments are set once and then used by all tasks in the DAG unless overridden.

Execution Sample

Apache Airflow

from datetime import datetime
from airflow import DAG

default_args = {
    'owner': 'airflow',
    'start_date': datetime(2024, 1, 1),
    'retries': 1
}

dag = DAG('example_dag', default_args=default_args, schedule_interval='@daily')

This code sets default arguments and creates a DAG that uses them for its tasks.

Process Table

Step	Action	Parameter Set	Value	Effect
1	Define default_args	owner	airflow	Sets task owner to airflow
2	Define default_args	start_date	2024-01-01	Tasks start from this date
3	Define default_args	retries	1	Tasks retry once on failure
4	Create DAG with default_args	default_args	default_args dict	DAG uses these defaults
5	Define task1 without override	owner	airflow	Inherits from default_args
6	Define task1 without override	retries	1	Inherits from default_args
7	Define task2 with override	retries	3	Overrides default retries
8	Run DAG	task1 retries	1	Task1 retries once
9	Run DAG	task2 retries	3	Task2 retries thrice
10	End	-	-	Execution complete

💡 All tasks run with parameters from default_args unless overridden explicitly.

Status Tracker

Variable	Start	After Step 5	After Step 7	Final
owner	undefined	airflow	airflow	airflow
start_date	undefined	2024-01-01	2024-01-01	2024-01-01
retries	undefined	1	3 (task2 override)	varies by task

Key Moments - 2 Insights

Why does task2 retry 3 times but task1 only once?

Because task2 explicitly overrides the 'retries' parameter (see execution_table step 7), while task1 uses the default from default_args (step 5).

What happens if a parameter is missing in default_args?

If a parameter is missing in default_args and not set in the task, Airflow uses its internal defaults or may raise an error depending on the parameter.

Visual Quiz - 3 Questions

Look at the execution table, what is the 'owner' value for task1 at step 5?

A admin

B airflow

C user

D undefined

💡 Hint

Check the 'Parameter Set' and 'Value' columns at step 5 in the execution_table.

At which step does task2 override the default 'retries' value?

A Step 5

B Step 6

C Step 7

D Step 8

💡 Hint

Look for the step mentioning 'Define task2 with override' in the execution_table.

If we remove 'retries' from default_args, what happens to task1 retries?

A Retries use Airflow default

B Retries become 3

C Retries become 0

D Task fails to run

💡 Hint

Refer to key_moments about missing parameters and Airflow's internal defaults.

Concept Snapshot

default_args is a dictionary of parameters set once for a DAG.
Tasks inherit these parameters unless overridden.
Common keys: owner, start_date, retries.
DAG uses default_args to apply consistent settings.
Overrides allow task-specific behavior.
This simplifies DAG management and reduces repetition.

Full Transcript

In Airflow, default_args is a dictionary that holds common parameters like owner, start_date, and retries. When you create a DAG, you pass default_args so all tasks inside inherit these settings automatically. This means you don't have to repeat the same parameters for every task. However, you can override any parameter for a specific task if needed. For example, if default_args sets retries to 1, but a task needs 3 retries, you override it in that task. This approach keeps your DAG clean and consistent. The execution table shows each step: defining default_args, creating the DAG, defining tasks with or without overrides, and running the DAG with the final parameters applied. Remember, if a parameter is missing in default_args and not set in a task, Airflow uses its own default values or may raise an error depending on the parameter.