🚀 go-pugleaf

   I just inherited a client this week who appears to have a long
history of data corruption on their server.  I'm still working on
the problem and doing all the maintenance stuff that their previous
consultants didn't do (like what?  oh, little things like performance
tuning, using current drivers, stuff like that).  From this
background, when the client tells me that they've observed
such-and-such, to some degree I have to take their word for it,
so take this article's information with a grain of salt.

   The server runs NetWare 3.12, 5 user, no patches, no upgrades,
(until yesterday) no tuning, nothing.  It's a Pentium-75 clone
with 16 MB of RAM, an ISA Ethernet card (sorry ... I don't recall
the make), and a PCI Adaptec host adapter with the AIC-7870
chipset.  The server is fed from an APC SmartUPS (I forget the
rating) which, thorugh PowerChute, indicates that their power is
quite clean (it has its own dedicated circuit and breaker) and
that the load is about 20% of what the UPS can handle.

   Workstations used to run VLM 1.10; they're now up to current
(from the 1.21 install kit, which seems to install 1.20), with
3C509 cards.  All workstations run DOS; one runs Windows
occasionally but that's not related to what's happening.

   Backups are done from a workstation using Cheyenne ArcSolo 2.2
for DOS through an Adaptec 152x card (using Cheyenne's driver,
not ASPI); I don't recall what tape drive they use.

   The only application they run is MAS90 accounting software.
Most of the time, everything works very nicely.  However, they
tell me that once in a while, files will vanish during the
backup or will go sparse.  From the backup logs, it does appear that
at least the vanishing occurs during the backup/compare process,
as ArcSolo will complain during the compare that a file is on
the tape but not on the drive.  No damage appears to occur to the
volume and no problems are noted in the system error log.  They
say that turning off compare after verify makes the problem go
away, though I don't know if that's true or if it just stops
them from noticing.

   They also tell me that if they use XCOPY with the verify option to
make backups to another directory, several files go sparse; if they
just use XCOPY without verifying, the files remain intact and
unmangled.  They have only tried this once so I don't know if
it's repeatable or semi-repeatable.

   I've looked at their workstations and server and can't see
anything too far out of the ordinary.  They will soon be getting
a collection of patches and upgrades for the server, and I'll
be replacing things like AIC7870.DSK version 1.00 with more recent
versions.  I'm not sure this will fix the problem, though, and
I don't 100% trust their troubleshooting skills.  Unfortunately,
the problem is intermittent and so unless the XCOPY trick works
repeatably I can't force it to happen before my eyes (I didn't
get to try this while I was there, as they only discovered it
after I had left).

   Any ideas?
-- 
stephen@bokonon.UUCP                     ...!{xrtll,gts.org}!bokonon!stephen
----------------------------------------------------------------------------
Stephen M. Dunn, CNE, ACE, Sr. Systems Analyst, United System Solutions Inc.
104 Carnforth Road, Toronto, ON, Canada M4A 2K7          (416) 750-7946 x251