During a recent batch of deposits (see case 00477041 from the Getty for more info) we figured out that, due to an error in our one of our input files that creates the METS, our original METS file did not have unique file IDs in the structMaps for the preservation master and modified master representations. It was simply repeating the same file ID from the access file representation.

For some reason, this did not cause a METS validation error when we validated against the Rosetta XSD before deposit, nor when it was deposited. However, it resulted in the METS in Rosetta not having any files listed in the structMaps that contained the duplicate files IDs. Since the fileIDs were correct in the fileSec it imported all the files properly, but they were not referenced from any structMap. Fixing this using the back office would have taken too much effort so we deleted the ca. 150 IEs and redid them from scratch.

Might you be able to enhance the METS XSD so that it requires file IDs to be unique in each structMap? I know it is against Rosetta's data model to refer to the same file ID from different representations with the same structMap type so it would make sense if the XSD enforced this requirement.