This might be somewhat outside the realm of "alpha-osf-managers", and
more appropriately in the domain of comp.periphs.scsi and/or Adaptec's
(or Seagate's) technical support gurus, but just in case ...
We're having problems involving a Seagate ST15150N co-existing with
other devices on a SCSI bus controlled by an Adaptec AHA-1742A. I'm
not claiming to be a wizard when it comes to hardware issues in general
or with SCSI problems in particular, but then again I'm not exactly in
the neophyte class, either. But I'm stumped. Here are some pertinent
details:
The Adaptec board is in an EISA slot on a DEC 2100/A500MP system (one
of their multi-CPU Alpha AXP systems). If the ST15150N is on the bus
alone, it works just fine. However, if one or more other devices are
attached to the same bus, they and/or the ST15150N are quite flaky (at
best), or refuse to work at all. (Typical symptom might be that a SCSI
operation is initiated, and the remote LED comes on and stays on, and
the operation never seems to complete.) Write operations seem to have
a much higher probability of failure than do read operations.
As long as the ST15150N is not connected, I can connect several other
SCSI devices to this bus simultaneously, and all can be used without
problem. But once the ST15150N is attached, attempts to use the other
devices have a high probability of failure, even if the ST15150N itself
isn't being used. Things, in fact, are flaky even if only there's only
one other device on the bus with the ST15150N.
As an example of our {success,failure} configurations ... What we'd
like to have on this bus are one each of {ST11950N,ST15230N,ST15150N}.
I can put the first two drives on the bus together with no problem at
all. If I put either one of them on the bus with the ST15150N, things
may or may not work for a while, but eventually fail. And if I put all
three drives on the bus together, all of them are generally unusable as
soon as the system is booted.
The ST15150N has also been observed to be solidly nonfunctional if it's
sharing this bus with nothing else besides a Digital DAT drive.
I've checked all the obvious things: All the devices have the internal
termination disabled, an external active terminator is installed on the
last device, and TERMPWR is present. (It's not quite at +5V; more like
4.85V or so. But this should be good enough, oughtn't it?) All of the
cables are quite short (two feet or so). The Adaptec controller board
has been installed correctly (assuming the installation instructions are
correct ;-). It's unlikely to be a bad cable; I've tried replacements
for all the cables involved in all these permutations.
The obvious suspects are probably the ST15150N and the Adaptec board.
But I tried replacing the Adaptec board (we have a spare; both it and
the other board are brand new); the problem didn't change. And we've
also had the ST15150N swapped for a new one (as well as the enclosure
and power supply); again, no joy.
It's worth pointing out that we have lots of ST15150N drives on many
of our other Alpha boxes, but they're older, single-CPU systems, with
a different SCSI controller (no idea whose; DEC's own, maybe). And
all of them co-exist quite happily with one, two, or even three other
drives. This Adaptec board, incidentally, controls the DEC 2100's
second SCSI bus. The first SCSI bus which comes automatically with
the system is (I believe) a different board from the Adaptec board
which is giving us the problems. (Oh by the way, the ST15150N does
seem to work fine on the primary SCSI bus, co-existing happily with
a couple of DEC RZ28 drives. However it does become unreliable if
our external DAT drive is added; but this may be a case of exceeding
the maximum SCSI cable length, given the fact that (according to the
DEC 2100 documentation) there's already quite a bit of internal cable
on the first SCSI bus, inside the box. Anyway, that's one more data
point.)
So there's something about the Adaptec board which is responsible for
the problem, at least in part. However, this board is able to support
all sorts of other devices, either alone or in combination, without any
problem at all. It's just plugging in that ST15150N which causes grief.
Oh yeah, one other thing perhaps worth pointing out: Although the board
is designed and manufactured by Adaptec, it's sold by Digital. In fact,
it's what they supply if you order a second SCSI bus for the 2100 boxes.
So should I be expecting DEC field service to play a role here? (Given
that the problems appear only when a non-DEC disk is installed, I guess
the answer is probably "no".)
Any clues, hints, suggestions, or whatever? Many thanks in advance ...
Mark Bartelt 416/978-5619
Canadian Institute for mark_at_cita.toronto.edu
Theoretical Astrophysics mark_at_cita.utoronto.ca
"Nur eine Waffel taugt!" -- Parsifal, in an Eggo commercial
Received on Thu Sep 21 1995 - 23:18:48 NZST