Why 2 parallel tasks have same pid?

Airflow
1

I just configured airflow.cfg to work with LocalExecutor :

    executor = LocalExecutor
    sql_alchemy_conn = postgresql+psycopg2://airflow:airflow@localhost:5432/airflow
    sql_engine_encoding = utf-8

after initializing the DB , I run the following DAG:

with DAG(dag_id='a_example_parallel_v', schedule_interval=None, start_date=days_ago(2),) as dag:

    def task1_func(ti):
        print(f"pid: ({[os.getgid()]}task1: print from task 1")

    def task2_func(ti):
        print(f"pid: ({[os.getgid()]}task1: print from task 2")

    def task3_func(ti):
        print(f"pid: ({[os.getgid()]}task1: print from task 3")

    def task4_func(ti):
        print(f"pid: ({[os.getgid()]}task1: print from task 4")


    task1 = PythonOperator(task_id='task1', python_callable=task1_func, provide_context=True)
    task2 = PythonOperator(task_id='task2', python_callable=task2_func, provide_context=True)
    task3 = PythonOperator(task_id='task3', python_callable=task3_func, provide_context=True)
    task4 = PythonOperator(task_id='task4', python_callable=task4_func, provide_context=True)


task1 >> task2
task1 >> task3
task2 >> task4
task3 >> task4

parallel

  • I checked the logs of task 2 & task 3 and they printed the same PID.
  • If task 2 & task 3 have same PID, it means that they are not running in parallel.

I tried to changed:

execute_tasks_new_python_interpreter

to True (and got same results)

How can I configure airflow, in order that task 2 and task 3 run in parallel ?