This question is rather three questions, one to people familiar with  
NetBSD's various low-level SCSI drivers (i.e. those in sys/dev/ic),  
second to people deeply familiar with SCSI, third to people operating  
a (NetBSD) machine with lots of SCSI disks attached to one parallel  
channel.

We operate a file server (on 4.0 beta2/amd64) which has, attached to  
one of its two SCSI channels, an external box equipped with eight SCA  
drives (plus four vacant slots). From time to time (something like  
once a week), ahd(4) spits out a frightening amount of output  
starting with
ahd1: transmission error detected
which I was told on this list was a SCSI parity error.

First question (to people familiar with NetBSD's SCSI ic drivers): Is  
ahd(4) just very verbose on this subject or would other ic drivers  
log (recovered) parity errors, too?

Second question (to SCSI experts): On an 160MTransfers/s (U320) LVD  
channel, how many parity errors would one be likely to expect?

Third question (to SCSI users): Is anyone operating a similar (or  
even larger) configuration, i.e. five to ten SCA disks in an external  
enclosing? If so: how often do you experience SCSI parity errors?

As I'm likely to be asked whether my SCSI bus was correctly  
terminated etc.: I've read the relevant parts (I hope) of spi4r10 and  
am confident that our setup is well within the limits (if the Chenbro  
backplanes, the Fujitsu discs, etc. are, of course). I've checked (by  
inspecting DIFFSENS) that the bus is indeed operating in LVD mode.

Thanks for any replies.