>nisimura@is.aist-nara.ac.jp (Toru Nishimura) wriets:
>
>Hi, and please forgive me about serial Emails.
>
>        [ Jason failed to boot 1.2G kernel with 5000/125] 
>
>I've run across yet-another-possiblity (during taking dinner).  Jason,
>your machine has an external housing holds RRD42 and RZ25, right?  My
>(might-sound-dumb) question is; if so is your SCSI cabling/terminatation
>configured properly?

same apologies, but:

This *has* caused several people real problems in the past.

If memory serves, Maximum Entropy had some nasty problems with initial
installs due to bad or marginal SCSI cables and poor termination --
even though Ultrix coped with it just fine.  Replacing the SCSI cables
solved that problem completely and permanently

This can be a particular problem with the 3MIN chassis (5000/125) with
the long cables that go under the CD-rom housing; they need
extra-special attention when installing or removing drives, and with
proper termination.

Also, in general NetBSD/pmax is just a _lot_ fussier (or less
tolerant) about marginal SCSI cabling than Ultrix.  We don't have the
support resources that DEC had in their hey-day, and in some respects
I think it's safer to not work at all on marginal configurations than
to rework the SCSI drivers from scratch to do better error recovery
(and risking  data corruption if we get it wrong.)

There _is_ a fix in the works for this, it involves moving away from
the old 4.4bsd SCSI driver code to the NetBSD machien-independent SCSI
code, which is better designed, cleaner, a lot more robust, and just a
whole lot more capable of dealing with SCSI problems.

That aside -- NetBSD/pmax with good SCSI cables and proper termination
seems to work flawlessy, with good performance (on ioasic machines
like the 3MIN) and with a wide range of drives.