Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watnot!watmath!clyde!cbatt!ucbvax!JPL-VLSI.ARPA!tencati
From: tencati@JPL-VLSI.ARPA.UUCP
Newsgroups: mod.computers.vax
Subject: Re: SBI Faults
Message-ID: <870303101634.069@Jpl-VLSI.ARPA>
Date: Tue, 3-Mar-87 13:16:34 EST
Article-I.D.: Jpl-VLSI.870303101634.069
Posted: Tue Mar  3 13:16:34 1987
Date-Received: Fri, 6-Mar-87 00:21:20 EST
Sender: daemon@ucbvax.BERKELEY.EDU
Distribution: world
Organization: The ARPA Internet
Lines: 35
Approved: info-vax@sri-kl.arpa

Nick,

Please promise you won't laugh..

I had a problem where my 780 would bugcheck about 3 times a week.  DEC
Field Service escalated it to the point where I had 2 District guys sitting
in the computer room waiting for the system to crash so they could run dumps
and look at stuff.

It was very interesting because the device that was causing the problem was
an RM03 that was spun down and had been for about a month.

Here's what the problem was:

Our computer room is extra-cold due to another computer room sharing the A/C 
with us.  It is also a water-cooled system as opposed to freon or some other 
chemical.  This caused the humidity to be higher than usual in the computer
room.  Over an extended period of time, this humidity caused microscopic MOLD
to grow on the gold plate of a couple pins on the Massbus adapter for the RM03.

This mold would then cause electrical variations in on the board which would
cause it to write a bogus value into memory.  VMS would come along and try
to execute the instruction...BUGCHECK...

It took PAINSTAKING diligence on the part of DEC.  They showed me the mold
and I believed...  They replaced the "moldy" board and my system worked
fine (until the next problem), but they had fixed the bugcheck problem.

So as weird as it may seem, have them clean the pins on your boards.  It 
may solve your problem.  It couldn't hurt in any case.

Good Luck,

Ron Tencati
System Mgr, JPL-VLSI.ARPA