Basic DAG file to test parallel execution on Airflow

MyDAG.py

# First, let's import all the basic Airflow modules that we will be using on this DAG file
from airflow.utils.dates import days_ago
from airflow.models import DAG
from airflow.operators.dummy_operator import DummyOperator
from airflow.operators.python_operator import PythonOperator

import time

# Then, define your "default_args" dict to store a few of our DAG arguments, notice that the use of this dict is a good practice for DAG writing since we won't be needing to set things manually later on
default_args = {
    "owner": "admin",
    "start_date": days_ago(1)
}

# Create the Python function that will be executed by the parallel tasks (which is basically a sleep of 2 seconds)
def sleep_time():
    time.sleep(2)

with DAG('MyDAG', schedule_interval='@daily', default_args=default_args, catchup=False) as dag:
    begin = DummyOperator(
        task_id='begin'
    )

    first_task = DummyOperator(
        task_id='first_task'
    )

    first_parallel_task = PythonOperator(
        task_id='first_parallel_task',
        python_callable=sleep_time
    )

    second_parallel_task = PythonOperator(
        task_id='second_parallel_task',
        python_callable=sleep_time
    )

    third_parallel_task = PythonOperator(
        task_id='third_parallel_task',
        python_callable=sleep_time
    )

    last_task = DummyOperator(
        task_id='last_task'
    )

    end = DummyOperator(
        task_id='end'
    )

    # Now, we set the dependencies downstream and, using a list of tasks, our parallel tasks dependency before last_task
    begin >> first_task >> [first_parallel_task, second_parallel_task, third_parallel_task] >> last_task >> end