medici@elbereth.rutgers.edu (Mark Medici) (11/01/90)
I am having some continuing problems with a NetWare/386 3.1a (6/13/90)
server. The problem is reported drive failure and shutdown while
performing massive write operations (such as backing-up one server to
another). I have two other NetWare/386 3.1a servers running is
similar configurations without problems. Here are the specifics:
Hardware:
IBM PS/2-80 (8580-041), with
16MHz 80386
16MB RAM
3.5" diskette drive A:
IBM PS/2 ESDI fixed disk controller
Core HC-310 300MB ESDI fixed drive 0
IBM 70MB ESDI fixed drive 1
IBM Token Ring Adapter II/a (4mbps)
Standard IBM VGA display circuit and 8512 monitor
System Software:
Novell NetWare/386 3.1a of 6/13/90
AUTOEXEC.BAT contains:
Prompt Server IML-Proto^G$P$G
SERVER
STARTUP.NCF contains:
load PS2ESDI slot=8
AUTOEXEC.NCF contains:
file server name IML-PROTO
ipx internal net 80061106
load TOKEN
bind IPX to TOKEN net=80061100
mount all
load REMOTE <password>
load RSPX
load PATCHMAN
load DELDIRFX
load FIXOPEN
track on
load MONITOR
DRIVE ALLOCATION:
Drive 0 (300MB) has a 2MB DOS bootable partition, and a 298MB
NetWare/386 partition. The NetWare partition is devoted to
volume SYS: and hot-fix is enabled.
Drive 1 (70MB) is devoted as a single NetWare/386 partition of
70MB. This partition is devoted to volume USER: and hot fix
is enabled.
PROBLEM:
When performing massive writes to the server, server will disconnect
target drive (usually SYS:, a 300MB Core ESDI on an IBM ESDI
controller) complaining about drive failure. Client sees Error
writing to Fat and other write failures. Errors on server scroll by
too fast to see the beginning of the problem. No messages are
recorded in any error logs, because SYS: gets shutdown.
If I try to run VREPAIR or DOWN and EXIT the server, I get the
following ABEND message:
>System halted Wednesday October 31, 1990 5:48:35pm
>Abend: Nonmaskable Interrupt (Parity Error) Processor Exception
>Parity Error was generated by the system board
> OS version Novell Netware 386 V3.10 Rev. A 6/13/90
> Running Process: Polling Process
>
> Stack E4 DB 00 00 88 D6 F1 FF 00 00 00 00 0E 00 00 00
> 02 00 00 00 05 00 00 00 00 00 00 00 04 00 00 00
> 45 72 72 6F 72 20 77 72 69 74 69 6E 67 20 46 41
Attempts to copy diagnostic image to diskette failed, regardless of
capacity diskette used.
A previously recorded ABEND of the same nature produced only a
slightly different stack listing:
> Stack D7 DB 00 00 00 00 00 00 00 00 00 00 0E 00 00 00
> 02 00 00 00 05 00 00 00 00 00 00 00 04 00 00 00
> 45 72 72 6F 72 20 77 72 69 74 69 6E 67 20 46 41
What I've tried:
I have tried two completely different boxes, different controllers,
different token-ring cards, different drives, keyboards, etc. The
only non-IBM part in the box is the Core HC-310 fixed drive, which is
NetWare/386 certified. This system initially had an 80387 math chip
installed. Removing the math chip (which was done between the first
and second stack listing) seemed to eliminated the problem. However,
the the system just abended again while I was backing up another
server on the same ring using NBACKUP. I have also reformatted the
drives, tried using them with and without running Surface Analysis
(they all passed), and have re-installed the complete NW386 3.1a
package at least 6 times. Nothing has eliminated the problem.
Plea for help:
What is really puzzling is that I have two other servers similarly
configured that are experiencing no such problems. Maybe this is just
because they don't normally get the amount of writes I've been doing
on this box.
I have tried contacting my Novell reseller, but they are slow to
respond because we did not purchase the hardware from them (2 - 3 year
old systems from an IBM grant). I have been hesitant to contact
Novell because of their outrageous technical support costs ($200/hr!).
If anyone can offer any advice or suggestions I will be most grateful.
------------------------------------------------------------------------------
Mark Medici/SysProg3 * Rutgers University/CS-USG * medici@elbereth.rutgers.edu
------------------------------------------------------------------------------
--
------------------------------------------------------------------------------
Mark Medici/SysProg3 * Rutgers University/CS-USG * medici@elbereth.rutgers.edu
------------------------------------------------------------------------------