Subject: DLT1 (SCSI) tape drive
To: None <email@example.com>
From: Lucio De Re <firstname.lastname@example.org>
Date: 05/13/2003 15:46:30
NetBSD 1.6 kernel with SCO/SVR4 emulation. On an Intel Tupelo server
(dual CPU) which has two SCSI adapters on board. All from memroy,
right now, as the host is waiting at a kernel debugger prompt.
I have an HP DLT tape drive as device 5 on the first, externally
acessible SCSI adapter, the only device, with a passive terminator.
It worked OK given a
# cd /
# find . | cpio -ovH newc | dd if=/dev/rst0 bs=64k
but trying to read the tape is not as happy. I used, having checked
the CPIO man page:
# cpio -iH newc --io-size 65536 < /dev/rst0 > /dev/null
which resulted in repeatable panics, the most recent:
panic ahc_action: not tagged and device busy
Stopped in pid 220 (cpio) at cpu_Debugger+0x4: leave
I can provide more details if there are no suggestions on what to
type at the db> prompt first.
The HP tape drive is in fact OEM, but the manufacturer's name
escapes me. It's not one I'm familiar with. HP call it a Surestore
DLT VS80. HP suggest using a blocking factor of 64K and it certainly
made a difference in writing performance, while I got I/O errors
trying to read with the default 512 bytes input block size.
It's important to get this working as it is our choice of backup
medium and, apparently, ought to interoperate with SCO Open Server
5 which is our primary server (sigh!) with NetBSD as the "hot
I'm not sure how much I could assist in identifying the problem
more accurately and fixing it, but I'm willing to try.
dmesg.boot and other details available on request. The kernel is
not a default, but should be fairly typical, I can supply the
"config" file too.
The Tupelo server has two ahc7899s (I seem to recall) on board.
The second one is used for the hot-swappable bays with a SAF-TE
backing plate ('scuse the terminology, I'm a little out of my