scsi tape problem - goes and won't come back without reboot

[Danny Yee]

I have a server which every so often "loses" its SCSI tape drive.
Dumps stop working a command like "mt rewind" returns an error
"/dev/tape: No such file or directory".

The message log shows
	Aug  3 04:56:40 mail kernel: scsi1:0:5:0: Attempting to abort cmd c1bda980: 0x0 0x0 0x0 0x0 0x0 0x0
	Aug  3 04:56:40 mail kernel: scsi1:0:5:0: Command found on device queue
	Aug  3 04:56:40 mail kernel: scsi: Device offlined - not ready after error recovery: host 1 channel 0 id 5 lun 0
	Aug  3 04:56:40 mail kernel: st0: Error 20000 (sugg. bt 0x0, driver bt 0x0, host bt 0x2).

If I unload and reload the st module, the message log shows
	Aug  5 16:41:19 mail kernel: st: Version 20041025, fixed bufsize 32768, s/g segs 256
	Aug  5 16:41:19 mail kernel: Attached scsi tape st0 at scsi1, channel 0, id 5, lun 0
	Aug  5 16:41:19 mail kernel: st0: try direct i/o: yes (alignment 512 B), max page reachable by HBA 1048575

Which looks like it's finding the drive fine.  But attempts to use
/dev/tape (or /dev/st0) still fail.

I have no other SCSI devices (my disks are on an i2o RAID card), but
I can't remove the aic79xx module for the SCSI bus the tape is on --
it reports "ERROR: Module aic79xx is in use" and I can't work out
what's using it.

I can fix the problem by rebooting, but it would be nicer if I didn't
have to.

Any suggestions?

Danny.
Network Administrator
School of Medical Sciences
University of Sydney