Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.3 4.3bsd-beta 6/6/85; site amdahl.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!gamma!epsilon!zeta!sabre!petrus!bellcore!decvax!decwrl!amdcad!amdahl!esf00
From: esf00@amdahl.UUCP (Elliott S. Frank)
Newsgroups: net.rumor
Subject: Re: Computer Horror Stories
Message-ID: <2880@amdahl.UUCP>
Date: Wed, 5-Mar-86 12:53:39 EST
Article-I.D.: amdahl.2880
Posted: Wed Mar  5 12:53:39 1986
Date-Received: Fri, 7-Mar-86 07:39:38 EST
References: <14700001@hplabsb.UUCP> <12121@ucbvax.BERKELEY.EDU> <551@tekcrl.UUCP>
Reply-To: esf00@amdahl.UUCP (Elliott S. Frank)
Organization: Society for Megalithic Iron
Lines: 30
Keywords: meltdown

[......................................urp!]

When I was working at the computer center at Columbia University, we
had a 360/91 [limited edition, heavily pipelined, fast floating point,
dtl(=VERY VERY HOT, VERY FAST) logic].  One day I noticed little bright
colored cloth flags attached to the top of the cpu and I asked one of
the resident FE's about them.

He said that they were added after an incident elsewhere with another
360/91. The /91 was water cooled -- it had a closed loop system that
took about 400 gallons of distilled water.

It seems that one evening, the water system on that machine sprang a
leak, and dumped the contents of the chilling system under the raised
floor. 400 gallons spread over a large machine room is not very deep,
so nobody noticed. This was before subfloor water detectors were common,
so no alarms were set off. The computer room had enough air conditioning
to handle late-sixites/early-seventies(=low thermal efficiency) computer
equipment. So the chilled air flow through the machine was enough to
keep the thermal sensors in the CPU from shutting it down.

The water finally reached some of the subfloor electrical connections,
and shorted them out.  Unfortunately, they were the connections for the
chillers. With no cold air, the ambient in the CPU started to rise.
Before the sensors in the CPU could shut the machine down, it literally
melted from its own internal heat.

After the machine was reconstructed, we (and all of the other /91's) got
the cloth flags -- the operators were to check them, and if they were
not waving in the breeze, shut the machine down first, and check later.