Just a suggestion, Sepuku. Install smartmontools on all your systems, and configure the smartd daemon to run short tests and long tests on a regular schedule. My systems all run the short test -- just an an electronics self-test -- daily, and the long test, reading the media, weekly.
When I get bad sectors, which happens once in a while, I take the drive out of service and run badblocks against either a single section of the drive or the entire drive, depending on symptoms. If the problem drive is part of a RAID array, the server keeps on running, and I merely return the drive to service. If it's not part of a RAID array, such as in a desktop, I schedule downtime, and use my full and incremental backups as required.
|