that's turning out to be the problem, indeed.
When I look into the database, many files have uneven and inconsistent file type identification, and this is not only with PDFs. Also ms Word documents are sometimes identified as application/octet-stream or sometimes even as application/force-download (this sounds quite new ...).
I'm trying to understand what goes wrong with the production site, because as usual, in my local development VMs everything works smoothly.
Will update this thread as I make some progress, stay tuned.
./sysadmin ./web developer
./stealing bandwidth off of the internet since 1996