If it were a perfect world, then your second scenario would work just fine. However... I've seen MANY instances on this forum of people losing data trying it, for some reason or another.
If you want to try your luck, here is how I would approach it, just for education's sake.
1. DISABLE the VM and Docker services in settings. If you still see a Docker and VM's tab in the GUI, you didn't do it right.
2. Set ALL shares to Cache: Yes. This is important, it works differently than how you imagined, turn on help on that page to see why.
3. Make sure nobody is writing anything to the server, and run mover.
4. After mover is complete, make sure there are NO files left showing on the cache drive. It should be completely empty.
5. Proceed with the FAQ procedure to replace a failed cache drive.
6. Verify the new cache volume is mountable with no errors, and shows the full capacity you expect.
7. Change the shares that need to permanently be on the cache pool to cache:prefer, and run the mover. Make sure any shares that you DON'T want to use the cache pool are set back to Cache:No.
8. Re-enable VM and Docker services.
If everything works with seamlessly with step 5 and 6, then you can be confident that a failed cache drive would be handled correctly. If not, then thank your lucky stars you moved all the data off first.