Run pyspark from another server on airflow

Hi every one,
I have my DAG on airflow 2.6.2 as:
from datetime import datetime, timedelta
from airflow import DAG
from airflow.providers.apache.spark.operators.spark_submit import SparkSubmitOperator

Define default arguments for the DAG

default_args = {
‘owner’: ‘airflow’,
‘depends_on_past’: False,
‘start_date’: datetime(2023, 8, 2),
‘email_on_failure’: False,
‘email_on_retry’: False,
‘retries’: 1,
‘retry_delay’: timedelta(minutes=5),

Create the DAG

dag = DAG(
schedule_interval=‘@daily’, # Set your desired schedule here
catchup=False # Set to False if you don’t want to backfill past runs

Define the SparkSubmitOperator to run your PySpark script

spark_submit_task = SparkSubmitOperator(
application=‘/home/test/’, # t
conn_id=‘spark_conn’, # Connection ID to the Spark cluster defined in Airflow UI

spark_conn: I created on airflowweb => my spark server (included pyspark), not the same server with airflow.
I tested /home/test/ on my spark server, it is done. now I want to try run it on airflow but when I run dag it shows
from airflow.providers.apache.spark.operators.spark_submit import SparkSubmitOperator
ModuleNotFoundError: No module named ‘airflow.providers.apache’

But when i check in my airflow-webserver container with statement:
pip freeze | grep apache-airflow-providers-apache-spark → and it shows:

I don’t understand my problem. Can anyone help me to solve that? Thank you in advance.


Welcome to the Astronomer Forum!

If I understand your issue correctly, you have tested your Spark code directly using spark-submit on your Spark server and want to now use SparkSubmitOperator to run in an Airflow environment.

Can you provide the following details:

  • How are you running Airflow, are you running on your local system using Astro CLI?
  • Does your requirements.txt contain the package apache-airflow-providers-apache-spark ?
  • Do you also see the error on the Airflow UI? Because if the package is not properly installed, then the UI will throw an error and you will not be able to run your DAG.
  • Did you restart your Airflow after adding the Spark requirement?

Also, can you provide your entire DAG code in a code block ? Refer to this page to format your code.