I have a working (validated with no errors) database that includes a single product system, it's underlying processes, the federal elementary flows list, and an impact assessment method. Separately, I have a JSON file with processes, the federal elementary flows list, and the same impact assessment method. A subset of these processes are also in the first database, with newer versions housed in the second database. This JSON is validated with no errors when it is imported to an empty database.
When I import to the first database and select "Update data sets with newer versions", I get roughly 1250 exchangeID errors. Inspecting using SQL doesn't reveal much, except that the errors might all be with product and waste flows, and there aren't issues with updating emissions or resources. The other confusing aspect is that even with the exchangeID errors, running an analysis still reveals the results I would expect, and the model graph looks like it is complete.
Is it possible that the errors are associated with flows or processes that don't exist in the destination database, and openLCA is surprised to see a blank ID when "update" was selected? The JSON I am attempting to import doesn't contain any sort of database ID, so this would sort of make sense.