I am currently using a Norton course pack as the basis for my question pool. However, I have been editing them to remove all the essay and matching-item questions from each chapter pool, reserving them in another category for future use. Example: for Chapter Seven, I remove all the essays and matching items and put them into a category called "Unused Chapter Seven," leaving only m/c and t/f items in "Chapter Seven" itself.
Today, something unusual popped up. In a short-semester class created largely by the Import Canvas Course (selected content), I noticed that the question banks had reverted back to their original forms. In the above example, "Chapter Seven" included all of the original questions, including the ones I had moved to "Unused Chapter Seven," even though I had NOT tagged that question bank for importation! I discovered that going back and importing the "Unused Chapter Seven" question bank into the new course fixed the problem and re-separated the questions.
Meantime, however, this bug caused me quite a bit of headache, as we were mid-exam when folks started emailing me about all the "extra" essay questions they had to answer. I ended up having to rebuild the exam from scratch and giving everyone an extension.
Could this have something to do with the metadata in the file? And would the path of least resistance going forward be to actually delete the unused questions in the first place?