https://bigdataschool.ru/blog/spark-pipelines-debugging-with-dagster-instead-airflow/