[comp.unix.i386] 386/ix 2.0.2 install help

bill@inebriae.UUCP (Bill Kennedy) (11/09/89)

I'm having a problem installing ISC 2.0.2 that I'm sure is something
others have seen before, sorry for the length, just `n' if you don't
want to hear about it.  Also, please email if you can help, I doubt
that this is a common or general problem.  In fairness to ISC tech
support, they have been helpful, but I seem to acquire the needed
wisdom incrementally (one call per step) and I need to buckle up and
get the darned thing done.

The system is an old Micronics 16MHz motherboard, 6Mb RAM, no math chip.
The BIOS is Award 2.09 (my vendor says that's OK for ESDI), monochrome
video, Everex 60Mb tape, Computone AT-8, Logitech bus mouse, vanilla
cloneware serial/parallel card.  Everything but the video card, disk
controller, and memory card were removed for the install.  The drives
are a pair of CDC 150Mb ESDI's. 967 cylinders, 9 heads, 36spt.  The
hard disk controller is a WD-1007-WA2 (hard/floppy combo), '89 copyright
on the BIOS EPROM The system is running 1.0.6 with a pair of ST-506 drives.
Here's what all transpired.

First attempt (after formatting both drives with the controller BIOS,
no alternate sector mode) stopped when the boot floppy tried to sign
on, it stalled (hard stop) trying to display the amount of memory
found.  ISC tech support suggested trying another monochrome card, I
snorted at that, but another video card did the trick.  I've been able
to use either card since, but in fairness, ISC was correct, I was just
being hard headed.

The next symptom was an inability to make a 32Mb DOS partition starting
on cylinder 1.  The fdisk would ask where I wanted it, I'd reply cylinder
1 for 199 cylinders and it would faithfully lay down a 200 cylinder
partition starting on zero.  At this point I still had the Computone AT-8
in the backplane, ISC suggested removing it, I did, but made no real
progress.  I deleted all partitions, tried using DOS fdisk, no behvaior
change, so I just allocated the whole drive to UNIX.  The final symptom
at this point was no boot from the hard disk, no sign on or anything, just
a flashing activity light, some seek chatter on drive zero and an occaisional
activity flash on drive one.  After I removed the DOS partition altogether
I would get a very brief message on the screen about boot being unable to
open something and the BIOS memory check would start again.  Back to Santa
Monica for more help.

A side note to ISC if you're reading this, I understand how busy the phones
get and hold on toll is an unavoidable evil, but a phone system that just
drops the connection after five or six minutes is infuriating.  It happened
four times in one day, it can't be an isolated problem.  When you're paying
prime time LD to sit on hold it's a real attitude challenge to be cheerful
when you're just dropped without ever having reached a human...

Third try at ISC got a pretty savvy guy.  The other two were OK, but this
was the first tech I could understand without effort and who seemed to have
a pretty good knowledge of the product.  He was also sympathetic to my
asking for his name so I wouldn't have to explain everything from scratch
on each call, I haven't been able to reach him since :-(  He suggested that
I reformat the drives and use the alternate sector mode.  The trade off of
losing one sector per track vs getting installed seemed reasonable, so I
did it.  I also used DOS' (3.3) fdisk to lay down the DOS partition.  All
seemed normal, the DOS partition formatted OK and seemed to work (stay
tuned) and 2.0.2 seemed to install OK, the controller had allocated the spare
sector for the bad spots on the drive.  This is where I stuck and abandoned
the install and went back to 1.0.6.

Now the system behaves as though it really wants to boot off of the hard
disk.  It comes up and signs on with "Loading the UNIX Operating System..."
and the drive zero activity light flashes a few more times, but the boot
does not proceed from there.  I'll speculate that the first level boot
sector reads in OK but the controller/drive aren't being handled correctly
from there and there isn't enough read in to be able to display a meaningful
complaint.  I thought it might be waiting for a <CR> or something, but it
doesn't time out and proceed, it just stalls.

With as many 2.0.2's out there installed and reading this group, I'm pretty
sure that one of you will be able to help me quicker and easier than my
finding that tech at ISC again.  I'm not bashing their tech support, I
must emphasize that each time I was able to get through I was given a
suggestion that got me one step farther, but I don't know how many steps
are remaining...  Surely someone has encountered this already.  Is it the
second ESDI?  At one point I did disconnect the other drive, but it didn't
seem to help or harm (and it's a pain to keep setting up and setting up
again).  I can't return to the effort until Saturday and I will try with
the second drive disabled, but eventually I have to have both working.  Has
anyone installed 2.0.2 with two ESDI's?  If need be, I'll get another BIOS,
but my vendor (who has never been mistaken before) assures me that Award
2.09 is old but still OK for a WD1007 controller.  If I do need another
BIOS, please tell me what to get (sorry, but only known-to-work please, I
don't want to add another level of distraction) and I'll do it.

Finally (mercifully!) some maybe related observations.  When it wouldn't
boot from the floppy I tried AT&T Vr3.1 with the exact same result, but
Microport V/AT 2.4 came up and played.  I didn't proceed with either of
the other OS' but the boot behavior might suggest something.  There are
also some anomalies in DOS' behavior.  The format seemed to go OK and the
restore from diskettes appeared normal, but chkdsk reports all manner of
things that should and shouldn't be directories and the files that I
restored don't seem to be there.  That suggests (to me) a BIOS problem,
suggestions and experiences are eagerly sought and most welcome.  Please
email to the .signature address, that's the system I'm trying to install,
it's back up on 1.0.6 (hopefully for only another day or so).  Thanks a
bunch, and sorry for the length.
-- 
Bill Kennedy    {texbell,att,cs.utexas.edu,sun!daver}!ssbn!bill
                bill@ssbn.WLK.COM  or attmail!ssbn!bill

bill@ssbn.WLK.COM (Bill Kennedy) (11/13/89)

In article <441@inebriae.UUCP>, bill@inebriae.UUCP (Bill Kennedy) writes:
> I'm having a problem installing ISC 2.0.2 that I'm sure is something
[ I said it was long... ]

Mostly I was describing a situation where 2.0.2 just would not behave
with a WD-1007-WA2 controller and a pair of 150Mb CDC drives.  Many
thanks to Don Ahn (donahn@gypsum.berkeley.edu), Jim Frost (madd@world.std.com),
A. Lester Buck (buck@siswat.UUCP), Eric Schnoebelen (eric@egsner.cirr.com),
and Randy Smith (rls!randy@tut.cis.ohio-state.edu) for their suggestions
which I will summarize.  The problem turned out to be in my museum piece
motherboard BIOS, Award 2.09.  I moved the controller and drives to a
system with a newer BIOS and both DOS and V/AT (the other machine is a
'286 so I couldn't try 2.0.2) work just fine.  I'm quite sure that ssbn
will too with a more up-to-date BIOS.

The suggestions were all tried and each had some effect.  Don Ahn
discussed two symptoms I had for "hang at boot", the first is where
the system just seeks and seeks but never boots meaning the controller
is configured wrong.  It can also mean that you formatted with the
install program rather than the controller BIOS.  The other is sign
on and die, that's what I got when I enabled alternate sectors.  This
means that the kernel is looking for something it can't find.  In my
case that was the hard disk controller!  The BIOS could wrench out the
primitive and second level boot, but could proceed no farther.

Jim Frost suggested bad spots in unfortunate places and one drive did
have a bad sector on cylinder 200, first one after the DOS partition :-(
The other didn't, it's now drive zero.  A. Lester Buck suggested using
the install formatter and this got me the responses that the documentation
suggested.  That was particularly helpful for entering bad spots the way
CDC reported them (cyl, head, bytes from index).  With the controller
BIOS formatted drive I had to specify absolute sector number (shudder).
Eric Schnoebelen suggested bringing up Microport V/AT to see if it would
play, that's what I eventually did to verify that the drive and controller
were OK.  Randy Smith cautioned me to use only the controller BIOS, he's
right, but it was interesting to see the different personalities and
failures albeit frustrating.

Karl Denninger (kark@ddsw1.mcs,com) had the most novel suggestion,
pitch 386/ix and stick with SCO.  That's not a terrible idea if you
just want to use the thing.  I need to run X-windows and ISC is worlds
ahead of SCO in the variety of displays supported.  My problem is/was
that I couldn't get the underpinning to work.  I'm expecting an update
to my list of OK VGA's and monitors which I will post in another week
or ten days.

The end of the tale is when you first run into trouble, try and have
another system available so that you can isolate the problem.  That's
probably no easier for most of us than it is in Pipe Creek, TX but I
was lucky enough to have a '286 around with a newer BIOS.  The other
thing to do if in doubt is (as Karl suggested) pay a dealer or VAR to
get it up and running.  If I calculate what I saved by doing it myself
(it's still not done) I'd make less than minimum wage...  Thanks again
for the help.
-- 
Bill Kennedy  usenet      {attctc,att,cs.utexas.edu,sun!daver}!ssbn!bill
              internet    bill@ssbn.WLK.COM   or attmail!ssbn!bill

cld@kd4nc.UUCP (Charles D'Englere) (11/17/89)

Bill, 
	There is a problem using two ESDI drives on the WD1007A-WA2 
controller. The controller will cause the system to lock-up the hard 
disk routine, you will still have control of the console but won't be 
able to do anything. The only solution to the problem was to get 
Western Digital to replace the faulty controller with one the new ones, 
ie. WD1007V-SE2. The new controller supports drives up to 15 mega bits
per second, 2048 tracks, and has full track buffering. This controller 
really flies on a Micronics 386-25. If you need any further information 
give me a call. Hope the information helps.

Charles


=============================================================================
| Charles L. D'Englere                 UUCP: {gatech}!kd4nc!charles         |
| 1409 Bronze Leaf Ct.                Phone: 404+3259074                    |
| Stone Mountain, GA 30083                                                  |
=============================================================================