Tried to break it, actually it does a very good job on resolving inconsistency issues. Only thing that could happen is two instances of the updater running the same time (e.g. when one doesn't close properly). This keeps the database locked for the other instance. The sql driver of the valid instance will then check if the database was untouched for 10 seconds and restore its contents to the last checkpoint (around every 20 seconds changed data gets commited to the disk). If for any reason the crashed instance is still updating the lock status, the database will not be restored. To solve this issue just restart the computer. I didn't manage to break it so bad it wasn't able to restore.
Will implement some better warning messages for different errors and an option to get the database to factory settings...
Last edited by Kaldran : 03-13-2008 at 06:21 PM.
|