When we configure an Airflow DAG, we set the default values of various properties, for example, the retry delay. What should we do when we want to overwrite the delay in an Airflow task? What if we’re going to use a different delay in every task?

Table of Contents

  1. Get Weekly AI Implementation Insights

In this case, we have to include the retry_delay parameter in the constructor of Airflow operators, which delay we want to change. Every Airflow operator supports this parameter because it is defined in the BaseOperator. For the same reason, it is easy to overlook its existence because Airflow documentation does not show the parameters inherited from base classes.

Get Weekly AI Implementation Insights

Join engineering leaders who receive my analysis of common AI production failures and how to prevent them. No fluff, just actionable techniques.

Get Weekly AI Implementation Insights

Join engineering leaders who receive my analysis of common AI production failures and how to prevent them. No fluff, just actionable techniques.

Older post

How to find the Hive partition closest to a given date

How to use Airflow to find the Hive partition closest to a given date

Newer post

How to get names of columns with missing values in PySpark

How to get the names of missing properties for every row in a PySpark Dataframe

Engineering leaders: Is your AI failing in production? Take the 10-minute assessment
>