Killing linux with write-protect tabs
One of our servers is backed up to a 35/70 DLT drive. The other week it started giving us problems, and the drive stopped responding. I obtained a new drive (second-hand ones are cheap), replaced the old one, and everything worked fine... for about a week.
Then the second drive stopped responding. Same symptoms as the first, not showing up on the controller, and all the lights down the left hand side remaining lit after power-up. I'll spare you the details, but after a bit of investigative work I had reason to believe that perhaps our SCSI controller was having difficulty, and not the drive itself.
So, I shell out for a new controller, a sparkly new AHA-29160. These usually sell for $100+ retail, but using the wonders of eBay I picked one up for $20 including express delivery.
Drop in the card, attach the drive, power-up... The DLT drive still shows its lights all down the left-hand side. Damn. Then I notice the SCSI card isn't being recognised during boot. Re-seat the card (it's big!), power cycle. Everything comes up fine.
Boot into single user mode. Linux detects the new card and drive. Drop in an old tape in write-protect mode. Yup, reads it fine. Old data is still there. Happy happy day!
Wanting to perform a backup immediately, and the tape in the drive being the next one due for being overwritten, I wonder if I can flip the write-protect tab while the tape is in the drive... I can!
Sure enough, the light on the DLT panel extinguishes to indicate the tape is available for writing. Sweet. Time to rewind the tape and back-up.
# mt rewind (scsi0:A:6:0): parity error detected in Data-in phase. SEQADDR(0x1a5) SCSIRATE(0x98) (scsi0:A:6:0): parity error detected in Data-in phase. SEQADDR(0x1a6) SCSIRATE(0x98) (scsi0:A:6:0): parity error detected in Data-in phase. SEQADDR(0x1a6) SCSIRATE(0x98) (scsi0:A:6:0): parity error detected in Data-in phase. SEQADDR(0x1a6) SCSIRATE(0x98)
Oh bugger.
I guess I really do need to eject the tape before changing its write-protect status.
