My boot messages are appended below.
  I have a 4 tape changer. I'm pretty sure that I did get good backups
at some time in the past, but the past six months things have been much
less good. 
  Suspecting the NCR driver, I put an old 1542CF into my system, and moved
my "narrow" SCSI devices to it, leaving two Ultra+Wide devices on my Symbios
NCR controller. I get the same problems: the tape drive complains that the
host attempted to do overlapping I/O while backing up.

  I turned off reselect (and on the NCR, tagged queuing) for that device.
Now the SCSI bus hangs, timing out several times. This happens even if
I run the backup in single user mode --- I enabled networking, since the 
first step in my backup script is to confirm that the networked systems
that need to be backed up are available.

  The last two attempts required a hard reboot to get out of. I've been
through the Seagate web pages (they bought Conner, who had bought Archive,
I finally figured out this morning) on the device, and it corresponds
to the data sheet I have. I've turned parity on, turned it off, etc..

  At the "dump" level, when the overlapping I/O complaint shows up, dump
says that I'm at the end of the tape. Nonsense I say: I've only put 463Mb
on a tape. Even a 90m tape in DDS-I does 1Gb per tape. 
  I'm rebuilding with 1.3.1 patches now, but I don't expect a lot of
differences. 

  I have a 850Mb IDE drive that I can install with a boot loader if there
is some software someone might recommend that confirm proper operation of
the drive.

  [The CDROM was not on the chain for most of the tests --- I just put it
in the case this morning while I had things apart. sd2 is my boot drive,
which replaces a 4Gb drive that bit the dust in March. I also found the de0
complaint curious. I had intended to pull my ep0 and enable a de0 that I
got awhile ago. I had to do that yesterday to free up an ISA slot for the
1542. It seems to operate just fine.]

Copyright (c) 1996, 1997 The NetBSD Foundation, Inc.  All rights reserved.
Copyright (c) 1982, 1986, 1989, 1991, 1993
    The Regents of the University of California.  All rights reserved.

NetBSD 1.3 (SSW) #2: Sun Apr 12 14:36:47 EDT 1998
    mcr@istari.sandelman.ottawa.on.ca:/j/netbsd/src/sys/arch/i386/compile/SSW
cpu0: family 5 model 1 step 4
cpu0: AMD K5 (586-class)
real mem  = 66711552
avail mem = 63410176
using 839 buffers containing 3436544 bytes of memory
mainbus0 (root)
pci0 at mainbus0 bus 0: configuration mode 1
pchb0 at pci0 dev 0 function 0
pchb0: Intel 82437VX System Controller (TVX) (rev. 0x02)
pcib0 at pci0 dev 7 function 0
pcib0: Intel 82371SB PCI-to-ISA Bridge (PIIX3) (rev. 0x01)
Intel 82371SB IDE Interface (PIIX3) (IDE mass storage, interface 0x80) at pci0 dev 7 function 1 not configured
Trident Microsystems TGUI 9660 (VGA display, revision 0xd3) at pci0 dev 17 function 0 not configured
ncr0 at pci0 dev 19 function 0: ncr 53c875 fast20 wide scsi
ncr0: interrupting at irq 12
ncr0: minsync=12, maxsync=137, maxoffs=16, 128 dwords burst, large dma fifo
ncr0: single-ended, open drain IRQ driver, using on-chip SRAM
ncr0: restart (scsi reset).
scsibus0 at ncr0: 16 targets
sd0 at scsibus0 targ 1 lun 0: <DEC, RZ28     (C) DEC, 442E> SCSI2 0/direct fixed
sd0: sd0(ncr0:1:0): WIDE SCSI (16 bit) enabled
sd0(ncr0:1:0): 20.0 MB/s (100 ns, offset 15)
2007MB, 3045 cyl, 16 head, 84 sec, 512 bytes/sect x 4110480 sectors
sd1 at scsibus0 targ 15 lun 0: <FUJITSU, M2952Q-512, 0147> SCSI2 0/direct fixed
sd1: sd1(ncr0:15:0): WIDE SCSI (16 bit) enabled
sd1(ncr0:15:0): 40.0 MB/s (50 ns, offset 15)
2291MB, 5714 cyl, 5 head, 164 sec, 512 bytes/sect x 4693462 sectors
de0 at pci0 dev 20 function 0
de0: interrupting at irq 11
de0: SMC 21041 [10Mb/s] pass 2.1
de0: address 00:e0:29:11:0e:72
isa0 at pcib0
(bha probe): bha_cmd, cmd/data port empty 13
aha0 at isa0 port 0x330-0x333 irq 10 drq 7
aha0: model AHA-1542CF, firmware C.0
aha0: unlocking mailbox interface
aha0: async, no parity
scsibus1 at aha0: 8 targets
sd2 at scsibus1 targ 0 lun 0: <MICROP, 1588-15MBSUN0669, SN0C> SCSI1 0/direct fixed
sd2: 639MB, 1632 cyl, 15 head, 53 sec, 512 bytes/sect x 1309735 sectors
cd0 at scsibus1 targ 4 lun 0: <TOSHIBA, CD-ROM XM-5401TA, 3605> SCSI2 5/cdrom removable
st0 at scsibus1 targ 6 lun 0: <ARCHIVE, Python 28849-XXX, 4.CM> SCSI2 1/sequential removable
st0: density code 0x24, 512-byte blocks, write-enabled
ch0 at scsibus1 targ 6 lun 1: <ARCHIVE, Python 28849-XXX, 4.CM> SCSI2 8/changer removable
ch0: 4 slots, 1 drive, 1 picker, 0 portals
com0 at isa0 port 0x3f8-0x3ff irq 4: ns16550a, working fifo
com1 at isa0 port 0x2f8-0x2ff irq 3: ns16550a, working fifo
ast0 at isa0 port 0x1a0-0x1bf irq 9
com3 at ast0 slave 0: ns16550a, working fifo
com4 at ast0 slave 1: ns16550a, working fifo
com5 at ast0 slave 2: ns16550a, working fifo
com6 at ast0 slave 3: ns16550a, working fifo
npx0 at isa0 port 0xf0-0xff: using exception 16
vt0 at isa0 port 0x60-0x6f irq 1
vt0: unknown trident, 80 col, color, 8 scr, mf2-kbd, [R3.32]
vt0: console
fdc0 at isa0 port 0x3f0-0x3f7 irq 6 drq 2
fd0 at fdc0 drive 0: 1.44MB, 80 cyl, 2 head, 18 sec
biomask 1440 netmask 1c40 ttymask 1c42
de0: enabling 10baseT port
boot device: sd2
root on sd2a dumps on sd2b
root file system type: ffs
de0: enabling 10baseT port
de0: abnormal interrupt: receive process stopped