We have many users using relatively small VM workstations.  OCR of imported documents is a significant burden on these machines so we advise these users to not "Generate Text" when they import.  Instead we run an overnight workflow that grabs today's un-OCRed documents and sends them to DCC.
  It is helpful to have this capability as an option.  However, I have noticed, especially with large old files, that a single failed page (a photo, a bad photocopy, etc...) will stop the OCR of the rest of the document.

  Is there no way to have the OCR run continue after a single failed page?