ChrisRose
Databricks Partner

The issue is with a calculated field using the first_value aggregate function invoked as a window function with the OVER clause - 

first_value(Customer_ID) OVER (PARTITION BY customer_name ORDER BY date(submitted) ASC) as Customer_ID.
 
The field calculates correctly in a notebook or SQL Editor, but when run in a job via notebook the value comes back null for a particular record. It does not happen for all records. Please let me know if that is enough information, or if you need more to go on.