ashb commented on issue #5615: [AIRFLOW-5035] Remove multiprocessing.Manager in-favour of Pipes URL: https://github.com/apache/airflow/pull/5615#issuecomment-515402559 I'm running a comparisson between 1.10.3, 1.10.4rc3 and this branch with the following dag: ``` from airflow import DAG from airflow.operators.dummy_operator import DummyOperator from airflow.operators.python_operator import BranchPythonOperator from airflow.operators.bash_operator import BashOperator from datetime import datetime, timedelta default_args = { 'owner': 'airflow', 'depends_on_past': False, 'email_on_failure': False, 'email_on_retry': False, 'retries': 1, 'retry_delay': timedelta(minutes=5), } def make_dag(id): dag = DAG( dag_id=id, start_date=datetime(2019, 7, 1), schedule_interval=timedelta(minutes=10), default_args=default_args, catchup=False, ) with dag: start = DummyOperator(task_id='start') bulk = [ BashOperator( task_id=f'print_date_{n+1}', bash_command='for i in 1 2 3 4 5 6 7 8 9 10; do sleep $[ ( $RANDOM % 10 ) + 1 ]s; date; done', ) for n in range(20) ] start >> bulk def should_run(**kwargs): print('------------- exec dttm = {} and minute = {}'. format(kwargs['execution_date'], kwargs['execution_date'].minute)) if kwargs['execution_date'].minute % 20 == 0: return "dummy_task_1" else: return "dummy_task_2" cond = BranchPythonOperator( task_id='condition', provide_context=True, python_callable=should_run, dag=dag, ) cond << bulk dummy_task_1 = DummyOperator(task_id='dummy_task_1', dag=dag) dummy_task_2 = DummyOperator(task_id='dummy_task_2', dag=dag) cond >> [dummy_task_1, dummy_task_2] return dag for n in range(20): id = f'dag_{n:02d}' globals()[id] = make_dag(id) ``` I should have numbers by the end of the day with any luck
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
