Trailblazer Setup for HDB <--[NO! NO! NO!] for the UNIX PC

[Bill Mayhew]

If my experience is typical, not letting out HDB is costing AT&T
more to support the 3B1.  I bought one ofthe fire-sale 3B1's back
in 1987.  Here is the hardware and O/S configuration:

3B1
2 meg main memory on the CPU board (P3..P5)
67 meg miniscribe winchester
rel 3.51
development set


When I got the machine, was disappointed that the machine would not
run for more than a day or two without crashing.  Usually, the
crash didn't even result in any sort of panic message or entry in
unix.log; just dead.  Only thing working was the mouse pointer
could still be moved around on the screen.

I complained vociferously to the hotline service, which I must
admit tried much harder than many other vendors to be helpful.  It
was difficult to get "the ticket escalated" beyond the first tier
of people at the hotline, however.

After quite a bit of finger pointing, I removed every stitch of my
own software from the system and let it run.  The crashes
continued.  They accused my Hayes modem, which I was using because
I suspected the on board modem (OBM) was problematic.  I removed
the Hayes modem from tty000 and used only th OBM.  Still more
crashing.

With 2 months of my 90 day warranty up, AT&T finally agreed that
the uucp implementation must be at fault.  A person at the hotline
agreed to uucp me a new copy of uucico, which was interesting,
since that was the trouble-maker software to start with.  The new
uucico finally arrived, and I mv'ed it over to proper directory.
More crashes.  With only a couple of weeks left in the 90 days AT&T
was convinced that the motherboard had to be the culprit, so they
sent a person out with a new motherboard.  It was their $$, so I
figured, "what the heck".  The new motherboard worked for about a
week, then the crashes showed up again.  I thought they had it with
the new board.

About a month after my warranty was up, a an archive file
mysteriously showed up in my uucppublic directory.  Upon
investigation, it turned out to be the HDB package.  I decided not
to ask questions about where it came from.  I unpacked it, and have
been using it over a year.  The crashes stopped completely the day
I installed the HDB.  Now it might just be coincidence, but I'm
convinced that the uucp that comes with the version 3.51 O/S
release is a complete piece of junk.  HDB running on my Unix PC has
been supporting heavy uucp traffic with both the OBM and a
trailblazer without incident.

I have no idea if my experience is typical, but if it is, if I were
the hotline, I'd find out where that HDB package is, and make sure
it is on that disk they're about to send out to fix the time zone
bug :-).  (I'm not holding my breath!)  The hotline could then make
the HDB basic networking utilities the one and only supported uucp
for all machines.  Surely supporting multiple uucp incarnations
must be more expensive than one?

Bill
wtm at impulse.UUCP