Having just dealt with something similar from Linux to windows, I’m 97% sure this is the cause. Mine was a “same file name” issue because capitalization.
Also, it could be a temp file or hidden file that’s created by a program, like the recovery files made by libreoffice/word. They sometimes use weird characters.
Seriously. I’ve often thought that we need a journal for failed studies and experiments, or at least have a journal that only publishes articles that have been confirmed by a third party. Make it worthwhile to publish failed results or results confirmed by another researcher. I’ve heard too many stories of grad students repeating tests ad nauseam until they get a result their PA likes. Publish or perish needs to be gotten rid of.
Hmm, 3tb to mirror all the architectures I have around is very tempting. I have an unused dual bay external, and another one that desperately needs (wants) to have its 4tb drives upgraded.