Can someone explain "Data Phase Error"

All,

Environment:
kernel - 2.6.8-24 (from SUSE 9.2)
tape drive - IBM LTO-1 (scsi-wide)
scsi controller - Adaptec 2940

Issue:
I am trying to build a pair of new Linux based tape servers.

I have written exactly one tape on each, I then swapped the tapes beween the units and tried to do a restore. One seems to be working fine on this first test, but with the other I get:

Dec 17 08:52:04 linux kernel: st0: Error with sense data: Info
fld=0x40000, ILI Current st0: sense = f0 2b
Dec 17 08:52:04 linux kernel: Additional sense: Data phase error

I would suspect a bad tape, but a simple "dd if=/dev/st0 of=/dev/null bs=256k" fails at a different number of blocks each time.

i.e. I have 130 GB of data on the tape, but I have not yet read back more than 6 GB of it. The data itself is unimportant, I'm just testing the process.

I am continueing to test and am now trying another tape, but I want to understand this initial failure if I can.

Thanks for any help.
Greg

Greg Freemyer

I believe I found the problem.

Somehow one of the servers was built with a Adaptec 2940 controller, and the other with a Adaptec 29160.

Since the 2940 does not support LVD (only SE), I'm pretty sure that is the issue.

I'm actually surprised that the 2940 based system worked so well. I successfully wrote out 300+GB of data and those tapes were restorable by the 29160-based system.

It was only the restores that were unreliable, and that was true regardless of where the tapes were written.

Is there a LINUX - SCSI troubleshooting guide that provides advice for the various SCSI kernel error mesages? Does it cover this situation?

Greg

Greg Freemyer