priority distress signal from AC 814 (WREN II HD)

[John C Slimick]

I am having immense problems adding a second hard drive
to a 3b2400 system. The 3b2 currently has 4 MB memory,
WREN II (72 MB), and SVR3.2. The second hard drive is
also a WREN II (72 MB).

Symptoms: After formatting the HD with the ATT tools
package, a two drive system is built using the "add second
drive" option in disk 1 of essential utilities. All goes well
for about 9 days. Then bad CRC records appear. I take system
to single user state ("init s") and repair ("/etc/hdefix -a").
First repair usually works. 24 to 48 hours later other bad blocks
appear. Repeat of single user and hdefix. This time we run out
of backup blocks to remap, hdefix throws in the towel, and
we reboot. However, nothing in the first 20 blocks on drive 1
can be found--sometimes track zero itself can't be found.    
Now fsck seems caught in a loop, and I power down by pulling the 
plug (the switch doesn't work at this point). 

I laboriously rebuild the system, but with little hope. If I reformat
drive 1 (the new one), the format program finds nothing to complain
about. The drive is under warranty and has gone back to Imprimis once.

Any hints? could hdefix have become pathological? I have not had
a single bad block on the original WREN II since we got the machine
1.75 years ago. One complication: the vendor in Columbus who sold
us the second WREN is out of business, at least six months.
If it were power supply, why don't the failures happen earlier?
And why just on one drive? If it is the ATT controller, why does it
fail on just the one drive?

email appreciated
slimick at unix.cis.pitt.edu
john slimick
university of pittsburgh at bradford
bradford pa
(814)362-7543