将其他参数传递给on

本文介绍了将其他参数传递给on_failure_callback的处理方法，对大家解决问题具有一定的参考价值，需要的朋友们下面随着小编来一起学习吧！ 问题描述

我想将其他参数传递给我的on_failure_callback函数，但似乎只需要上下文。如何将其他参数传递给该函数...尤其是因为我想在一个单独的模块中定义该函数，以便可以在我的所有DAGS中使用它。

I'd like to pass other arguments to my on_failure_callback function but it only seems to want "context". How do I pass other arguments to that function...especially since I'd like to define that function in a separate module so it can be used in all my DAGS.

我当前的default_args如下：

My current default_args looks like this:

default_args = { 'owner': 'Me', 'depends_on_past': True, 'start_date': datetime(2016,01,01), 'email': ['me@me'], 'email_on_failure': False, 'email_on_retry': False, 'retries': 1, 'retry_delay': timedelta(minutes=1), 'on_failure_callback': notify_failure, 'max_active_runs': 1 }

如果我尝试类似这种方式，则会抱怨：

If I try something like this airflow complains:

default_args = { 'owner': 'Me', 'depends_on_past': True, 'start_date': datetime(2016,01,01), 'email': ['me@me'], 'email_on_failure': False, 'email_on_retry': False, 'retries': 1, 'retry_delay': timedelta(minutes=1), 'on_failure_callback': notify_failure(context,arg1,arg2), 'max_active_runs': 1 }

所以不确定如何传递arg1和arg我想在一个单独的模块中定义的我的notify_failure函数2可以简单地导入我的DAG中

so not sure how to pass arg1 and arg2 to my notify_failure fuction that I would like to define in a separate module that I can simply import into my DAG

推荐答案

您可以在DAG级别定义args，然后可以使用partials包。即：

Assuming the args are something you can define at the DAG level, then you can use the partials package. ie:

from functools import partial def generic_failure(arg1, arg2, context): # do whatever default_args = { 'owner': 'Me', 'depends_on_past': True, 'start_date': datetime(2016,01,01), 'email': ['me@me'], 'email_on_failure': False, 'email_on_retry': False, 'retries': 1, 'retry_delay': timedelta(minutes=1), 'on_failure_callback': partial(generic_failure, arg1, arg2), 'max_active_runs': 1 }

调用 partial（generic_failure，arg1，arg2）将返回一个函数，但预计 generic_failure 中仍有许多参数，在上例中，该参数只是单个参数 context

Calling partial(generic_failure, arg1, arg2) will return a function expecting however many arguments are remaining in generic_failure, which in the above example is just the single param context