non-reproducible system problems: losing files & directories

[Pablo Sanchez]

Hi there,

	We're experiencing a very strange phenomena on our machine.  I cannot
replicate the problem, however, the problem does occur with enough frequency
to be a real pain in the butt.  

	During our normal work day (we do ACCELL/SQL, ACCELL/IDS application
development) all of a sudden a file will disappear or the system will reboot
itself.  Also, when the system reboots itself, a whole bunch of files will get
munged.  For instance our ``mview'' library one time took a hit.  Or a C
archive will.  What's strange is that the files (or directories) that take the
hit can be read-only (such as the mview library or the C archive) or can be
read/write files (such as file.db from my ACCELL/SQL application).  By the
way, by taking-a-hit I mean that the i-nodes are stepped on.  I'll do an
``ls'' on the file and the file size is something like 2 gazzillion bytes or
the permissions are all screwy.  Although in the case with ``file.db'', that
baby just disappeared.

	The thing is that when files get munged, it does not always correlate
to the system rebooting.  Also, when we reboot the system we may get an error
message as follows: "Hard disk controller failure".  The problem is not
centralized to one disk either (see below for our configuration).

	I apologize if this note seems to ramble because I cannot see any
logical way to present these crazy facts.  Anyway, any help or pointers would
be greatly appreciated.


Configuration:
--------------

Machine				Everex 386/25
Primary Drive			Maxtor model 8760E
Secondary Drive			Micropolis 1664-7
Specialix SI Intelligent I/O driver
SmartCache (tm) on the disk controller
Main Memory			16MB
Cache on Main Memory		??MB or K?
-- 

Pablo Sanchez                   ``Can I touch you to see if you're real...''
Transact Software, Inc                   - Payolas