Xref: utzoo unix-pc.general:4954 comp.sys.att:8957
Path: utzoo!attcan!uunet!samsung!brutus.cs.uiuc.edu!apple!ames!zorch!atari!portal!portal!cup.portal.com!thad
From: thad@cup.portal.com (Thad P Floryan)
Newsgroups: unix-pc.general,comp.sys.att
Subject: Re: 3.51m hangs (wish it didn't)
Message-ID: <27612@cup.portal.com>
Date: 6 Mar 90 10:27:13 GMT
References: <4230@cayman.COM>
Organization: The Portal System (TM)
Lines: 92

peter@cuba.cayman.com (Peter Schmidt) in <4230@cayman.COM> writes:

	Since installing 3.51m, I've had over 1 hang per week.  The symptoms
	vary slightly.  Sometimes the console and tty000 are merely locked up,
	and sometimes hitting a key on the console causes a sort of "spray" of
	garbage characters in a pretty exponential down the console screen
	prior to the lock up, like this:
	...
	On one occasion, both machines croaked on the same evening, with
	winter locking up both our phone lines (this did not enhance my
	popularity with my wife, since we didn't notice for some time).

	Now, I used to have hangs/crashes like this about once a month with
	3.51 on winter, often when starting up gnu emacs, and I was *hoping*
	3.51m would make them go away.  Can anyone shed light on this?

Some people in this area report similar problems with their systems, with or
without having upgraded to 3.51m.  Through a series of "20 Questions", I've
come to the conclusion that computers WITHOUT proper AC power protection have
problems, and computers WITH proper protection have no (or very few)
problems.  For the record, my systems have NO problems whatsover.  In fact:

	ksh 7523/7524> uname -a
	UNIX thadlabs SYSTEM5 3.51a mc68k
	ksh 7523/7524> date
	Tue Mar  6 01:41:40 PST 1990
	ksh 7523/7524> who -b
	   .       system boot  Jan  2 04:58
	ksh 7523/7524> who -r
	   .       run-level 2  Jan  2 04:58    2    0    S
	ksh 7523/7524> 

And the only reason I rebooted then was to format some tapes.  The system
whose data is shown above has been operating continuously since 1987, 24
hours a day, with only 2 interruptions, and both those were long-term (over
30 minute) area-wide power failures.  My other systems are running 3.51m and
have been operating since late January, again with no problems.  And I use
these systems VERY heavily (including GNU Emacs :-)

A bit over 5 years ago I rented a recording AC power monitor.  The crap it
recorded on the AC power line was simply incredible.  Voltage spikes averaging
up to 2,500 volts, hash, sine-wave distortion, even loss of an AC cycle or two
every now and then.  By correlating those events with environmental observation
s
(here in my home and lab), I noticed direct correspondences to:

	+ turning on/off flourescent lamps
	+ refrigerator or freezer motor(s) or compressors kicking in
	+ forced-air heater motor starting up
	+ use of drill and saw motor(s) in my garage workshop
	+ electrical storm activity
	+ turning modems on or off (i.e. switching off their power supplies)
	+ phase of the moon (??) (e.g. bad power coming in, perhaps from
	  other activities in the neighborhood)

The reason I rented that power meter was to locate WHY my other computers were
having strange and intermittent problems (e.g. floppies getting corrupted,
unexplained system crashes, etc.).

The solution to MOST those problems was to get a PROPER surge protector; the
one I selected (best available at that time and probably still is) was mfd by
GTE.  They cost only ~$50 then, new.  The use of those protectors IMMEDIATELY
stopped the "bad" floppy problem, the unexplained system crashes, "random" HD
file corruption, etc.

Since then, I've switched to using SPS (standby power systems (mfd by SAFE,
their Models 500 and 1200A)) and now my systems also ride through total
neighborhood power failures and are immune to under- and over-voltage
conditions.

Power-line perturbations can affect systems in strange and "mysterious" ways.
Poor AC power line grounding only aggravates the situation.  It's my firm
belief that AC power conditioning is MANDATORY if you expect reliable computer
system operation at home or office.  I also have humongous Topaz power-line
conditioners operating on my office computers (large mainframes) and they, too,
operate reliably now.

Another thing to look at: check if the humidity is too low where you operate
your computer(s).  Low-humidity conditions are conducive to static buildup
which can be hazardous to one's system(s).  I've seen terminals at my office
go bonkers simply by someone walking across their office floor (carpeted),
sitting down at their chair, and touching the keyboard with their hands; the
solution to this was to HEAVILY spray the carpets with an anti-static
material and to assure relative humidity (RH) is around 50%.  In homes with
forced-air heating, the RH can be quite low and also cause sporadic arcing
inside CRT enclosures ... the voltage anomalies induced in chassis by such
phenomena are also the cause of computer systems operating strangely. 

I'm convinced that 3.51m is reliable and that many of the problems people are
reporting are due to "outside" influences.

Thad Floryan [ thad@cup.portal.com (OR) ..!sun!portal!cup.portal.com!thad ]