I have two Intel boxes running NetBSD 1.2.1 acting as file servers.
The boxes have the same setup with an ATX motherboard, EM960 ethernet
card and two SCSI controlers.

One of the SCSI controlers is an Adaptec 2940 and has the Root disk
(A segate ST-51080N 1gig) and a HP DDS 3 dat drive. I use the segate's
termination options to terminate the end of the 50 way IDC used for
the internal SCSI bus the dat drive is connected to an IDC connector
about 2 inches before the end connector.

The send SCSI controller is an adaptec 2940uw which has a ~9gig data
disk connected to it. This disk is split into two partions a bit over
4gig in size

I have had no problems with this setup except when tying to restore
backups from tape. The backups are made with dump but a "restoe vi"
to inspect them usually find that the tape has checksum erros.
The backups are made overnight when things are quite and consist of
a level 0 dump of each of the data disk partions.

Any advice on how to pin down this fault would be helpful. I have the
suspecion that the SCSI bus is not properly terminated or there is a
bug in dump .

--Sean