Date: Tue, 23 Mar 2010 19:14:13 -0500
From: John Foust <jfoust at threedee.com>
Subject: Re: Leaving computers on... (was Re: Disc analyser news
update)
To: <cctalk at classiccmp.org>
Message-ID: <201003240014.o2O0EISH064655 at billY.EZWIND.NET>
Content-Type: text/plain; charset="us-ascii"
At 07:06 PM 3/23/2010, Dave McGuire wrote:
In my experience, on/off cycles kills equipment
much faster than
long "on" times.
Is that biased by the experience of failures that appear when
equipment is powered-on?
I offer you this unscientific anecdote:
Yesterday, I rebooted an older Xeon box with 42 days' uptime to apply
updates, but it decided it didn't want to come back up afterward: no
POST, subsequent poweroff. I troubleshot briefly, and determined that
after working perfectly for over a month at a stretch (and 4 years
before that), it now refuses to POST unless you have half - any half -
of the RAM in the first two slots. No obvious signs of magic smoke
leakage. Coincidence? I suppose it's possible. I kinda wish I'd just
kept the old kernel, in any event. But with the price of energy in the
Netherlands, I wouldn't leave a system like that powered on for any
length of time if my employer weren't footing the bill for it!
Is there anybody here who keeps classic big(-ish) iron running 24x7?
I've been told a story of a burning PDP-11/34 which has somewhat put
me off any notion of leaving my -8s turned on when I'm not within
reach of a killswitch & fire extinguisher. :)
-js