Here’s a picture of my pipeline, I’m using a CSV file from a 3rd party system as the new file and then I’m using a select statement from my database as the original document. I run them both through a mapper to ensure that the column headings both match and the sort is on the unique ID. I’ve also run the pipeline normally and in validation mode. For some reason 2 records keep coming through as insertions even though the unique ID exists in both files. All the other files flow correctly to either deleted or unmodified (there were other modified records from before, but they all properly flowed to the update output when I ran the pipeline normally earlier which can be seen here
So something about these 2 records is making the diff snap think they are new when they really are not.
Isolate just those 2 IDs, from the incoming data and from the original (Select). Validate and download the jsons from both streams, and compare those with Notepad++ or VSCode whatever you’re using.
