Hubert-Dudek
Databricks MVP

It will not have data in the destination during the first insert, so that it will execute .whenNotMatchedInsertAll() for every record. Also, when two new records arrive at once (with the same id and state) in the next upserts, it will insert both. For sure, what you need is to aggregate data before inserting ('attom_id,' 'state_code',MAX( 'sell_date').


My blog: https://databrickster.medium.com/