The current backup algorithm has a few problems:
- It fails to delete some files, which will confuse the system in the event of a restore (particularly segments).
- When changing storage volumes, it does not check to see if a file already exists before copying over it (even if it is identical).
- When copying to a new backup volume, all Aeon threads are busy, so everything on the server stops.
At minimum we need a process to mirror the backup volume (including deleting files as appropriate).