成功完成后,Airflow会多次重新运行一个任务

时间:2019-02-18 22:25:36

标签: airflow

按顺序重新运行任务(A)3次的最佳方法是什么?:

即任务A->任务A->任务A->任务B

我问是因为我将运行另一个单独的数据验证任务(B),该任务将比较这3次单独运行的数据。

所以这是我到目前为止所做的:

dag = DAG("hello_world_0", description="Starting tutorial", schedule_interval='* * * * *',
          start_date=datetime(2019, 1, 1),
          catchup=False)

data_pull_1 = BashOperator(task_id='attempt_1', bash_command='echo "Hello World - 1!"',dag=dag)
data_pull_2 = BashOperator(task_id='attempt_2', bash_command='echo "Hello World - 2!"',dag=dag)
data_pull_3 = BashOperator(task_id='attempt_3', bash_command='echo "Hello World - 3!"',dag=dag)

data_validation = BashOperator(task_id='data_validation', bash_command='echo "Data Validation!"',dag=dag)


data_pull_1 >> data_pull_2 >> data_pull_3 >> data_validation

这可能有用,但是还有更优雅的方法吗?

1 个答案:

答案 0 :(得分:1)

您可以尝试以下实现,我们使用for循环创建3个操作

from datetime import datetime

from airflow import DAG
from airflow.operators.bash_operator import BashOperator

dag = DAG(
    "hello_world_0",
    description="Starting tutorial",
    schedule_interval=None,
    start_date=datetime(2019, 1, 1),
    catchup=False
)

chain_operators = []
max_attempt = 3
for attempt in range(max_attempt):
    data_pull = BashOperator(
        task_id='attempt_{}'.format(attempt),
        bash_command='echo "Hello World - {}!"'.format(attempt),
        dag=dag
    )
    chain_operators.append(data_pull)

data_validation = BashOperator(task_id='data_validation', bash_command='echo "Data Validation!"', dag=dag)
chain_operators.append(data_validation)

# Add downstream
for i,val in enumerate(chain_operators[:-1]):
    val.set_downstream(chain_operators[i+1])

我将schedule_interval更改为“无”,因为使用'* * * * *'时,作业将被连续触发