The only one was to increase the "spark.databricks.execution.timeout" because the query needed more than 2.5h unfortunately