I thought I was being clever and made a snapshot of an image I was working on before making some DNS modifications. The modifications broke the entire app and I ended up having to restore it, but having never worked with EC2 snapshots before the process was a little more confusing than it is for vSphere or Virtualbox.
What I did was detach the current volume from the instance and renamed the volume to “(Whatever) – Broken.” Then, I created a new volume of the same size and used the snapshot ID of the safety snapshot I took previously. This created a volume, using the contents of my backup snapshot.
The problem came when I went to attach the new volume to my instance. When asked to choose a device to attach the volume to, the console strongly insinuates that /dev/sda maps to /dev/sdf, or that /dev/sdf should be selected if it is the root drive. So, I plugged /dev/sdf in as the device in question.
Needless to say, it broke, and gave the semi-helpful message “Instance does not have a volume attached at root (/dev/sda1).”
So, it seems snapshots are backups of individual partitions, not entire drives, and thus plugging /dev/sda1 into the device field fixed it.