I’m looking at the DB Merge when connecting to a SQL database. It looks like when “Fail on error” option is disable the merge doesn’t apply updates up to the row that errors. After that the actual data is updated. So if I set Batch size to 5 and there is only error in row 7, no actual updates will be performed up to row 10, then the rest of the data is actually updated. Even if batch size is set to 1. It looks like the Merge doesn’t apply anything until after the rows that errors. I would have thought that only the rows in the same batch should be rolled back.
Can anyone confirm if this is the intended functionality of the DB Merge node? It doesn’t seem logical. If batch sizes are set to 5, only the 5 rows in the batch that had errors, should be rolled back or ignored.