NetNews Usenet Archive 1993 #3/NN_1993_3.iso/spool/bit/listserv/banyanl/6031

Text File | 1993-01-28 | 5.7 KB | 95 lines

Comments: Gated by [e-mail]
Path: sparky!uunet!uvaarpa!darwin.sura.net!sgiblab!spool.mu.edu!howland.reston.ans.net!paladin.american.edu!auvm!VINES.COLOSTATE.EDU!DEANWALLACE1
Return-Path: <@VM1.CC.UAKRON.EDU,
             @VM1.CC.UAKRON.EDU:[e-mail]>
Message-ID: <[e-mail]>
Newsgroups: bit.listserv.banyan-l
Date:         Thu, 28 Jan 1993 08:44:05 MST
Sender:       Banyan Networks Discussion List <[e-mail]>
From:         [e-mail]
Subject:      re: Third Hard Disk
Lines: 82

Frank,
        We have several servers with more than two disks.
1. Most ESDI/IDE setups can handle only two. We have several WD1007Vs, and the
controller itself can handle only two disks.
2. I have successfully mixed ESDI and SCSI several times. The ESDI disks are
always disk1 (and disk2, if present). Then, you start with the SCSI ones. I
have done disk1 ESDI and disk2/disk3 SCSI, as well as disk1/disk2 ESDI and
disk3... SCSI. One of our servers has two ESDI (330 Micropolis, 660 Fujitsu
M2263), plus one SCSI (1.2 Fujitsu M2266) and plans to soon add a second
M2266. The 1542 also is connected to a DAT drive.

You should be able to add a SCSI controller such as an Adaptec 1542B and
connect a disk to it and have it appear as disk3. In such a case, I always get
the tape drive onto the SCSI controller also; if the previous tape controller
was an IB02B, the tape then runs a lot faster. In this case, where you have
two previous IDE (or ESDI) disks, you would configure the 1542 to not use its
BIOS; actually, you don't need to use the 1542 BIOS unless you are booting
from a SCSI disk - VINES loads its own 32-bit SCSI driver and ignores the 1542
BIOS after booting.

I recently had some painful experieces with SCSI setup, etc. I talked to tech
support at both Adaptec and Fujitsu, who both suggested that jumpers needed to
be changed on both the 1542 and the Fujitsu disks - this is on a server with 2
660 Fujitsu M2263s, one 1.2 Fujitsu M2266, plus an Exabyte 8200 8mm tape
drive. The jumpers I changed were for synchronous negotiation, and for the
termination power.

On the 1542, jumper 1 on J5 was installed; this seems to
clue the 1542 to request to use synchronous transfer mode the first time it
talks to any SCSI device. The device is free to agree or disagree; synchronous
mode seems to imply that it will keep devices from responding asynchronously
and treading on each other's transmissions. In fact, that's what we were
seeing - when disk3 was connected, VINES displayed a continuous set of errors
on disk2. When disk3 was then disconnected, disk2 showed no errors at all.
I have the impression that, after the initial negotiation, the controller and
the device remember to use either synch or asynch transfer. On the Fujitsu
then, I had to remove jumper 15-16 on CNH2, to tell the drive that the
'synchronous mode data transfer request' would be 'initiated by Host' (host =
host controller = 1542).

I think that the more important jumper was for the termination power. There is
evidently one of the 25 wires in the SCSI cable called TERMPWR, that is used
to power the terminators at either end of the SCSI bus cable. The 1542 always
puts power onto this wire. Other devices can or cannot also put power onto it.
The Adaptec manual says 'No more than five SCSI devices should be configured
to supply terminator power to a single SCSI bus.' I rather think that if
multiple devices are attempting to power the same wire, that strange things
will happen, since the power from different devices is bound to be slightly
different. Thus, on the Fujitsu drives, I removed jumper 1-2 from CNH4. Then,
if a terminator is on that drive, it is powered only by the drive itself;
the drive does not use the TERMPWR wire for power, nor does the drive supply
power to the TEMRPWR wire. Exabyte says that the 8200 does not supply power to
the TERMPWR wire.

Before I changed these jumpers, I had the three Fujitsu drives connected, with
factory defaults on all drives and on the Adaptec. I also had an external
Archive/Banyan 150 tape drive. About once a week, usually during a tape
backup, the computer would totally lockup, with the 'hard drive' light on (the
one that's connected to the 1542 to show the controller is selected/busy).
Since then, it hasn't happened. (Other things have, but not that...)

By the way, I want to compliment Exabyte for their technical support. We had
an 8200 that originally came from IBM, for our RS/6000. We wanted to recyle it
onto this server. We hooked it up, and it wouldn't work. I called Exabyte and
talked to Dennis, who does tech support for the Rocky MOuntain Region. He told
me that IBM does not use standard SCSI 'language' - there are specialized ROMs
in that drive - and it won't work on a regular PC. (We have successfully used
regular PC 8200 drives on an RS/6000 - you just have to tell AIX that it's an
'other' device.) He GAVE me new ROMs (no charge) and even volunteered to send
them 2-day FedEx. We talked a little more, and, since Exabyte is only about an
hour away, it turned out that one of their employees lives here, so Dennis
sent the ROMs home with this employee, who brought them TO MY HOUSE that
evening. We put in the ROMs (they came with a 25-page document detailing the
installation), and have had no trouble since.

Dean Wallace                                [e-mail]
Academic Computing and Networking Services  303/491-7444
Colorado State University                   FAX:[phone]
Fort Collins  CO  80523
   *Even if you're on the right track, you'll get run over if you
    just sit there.  - Will Rogers
   *Better to ask twice than to lose your way once.  - Danish proverb