We're running SLES 10 SP3/OES2 SP2 on a Dell server with an IBM LTO-4 tape drive and Arkeia Network Backup software. Sometimes our backups fail because the tape drive, /dev/st0, has disappeared from the system. The server may run for weeks or just days before it happens - we can't find any pattern. The only solution we've found is to reboot the server. Then /dev/st0 is visible again, and the tape drive works fine.

I've called Dell about this multiple times over the last two years. They've replaced the tape drive, and we install the latest firmwares as they become available. Dell doesn't seem to know what to do any more, figuring it's an OS issue.

Any suggestions on how to (1) troubleshoot the problem, (2) prevent it, or (3) fix the problem when it occurs without needing to reboot?