Date: Thu, 19 Jan 1995 02:04:02 +0000 (MST) From: Jeremy Chatfield <jdc@crab.xinside.com> To: kuku@gilberto.physik.rwth-aachen.de Cc: freebsd-hackers@freefall.cdrom.com Subject: Re: is it hardware? Message-ID: <199501190904.CAA14730@crab.xinside.com> In-Reply-To: <199501190936.KAA25561@gilberto.physik.rwth-aachen.de> from "Christoph Kukulies" at Jan 19, 95 10:36:58 am
next in thread | previous in thread | raw e-mail | index | archive | help
Christoph Kukulies writes:
> > Some of you may be aware that our Server is configured to send email,
> > by default, with a human readable stack trace and the configuration
> > file in use... We have received the email from this event. There is
>
> Now I'm aware, too :-) How many mails does Xinside, Inc. receive that
> way daily? ;-)
The quantity varies, significantly. BSDI have a contract with us to
supply the X Server for BSD/OS. BSD/OS 1.1 was missing some kernel
support for the X Server. Fail to install the patch that we
supplied, and the X Server will die at OSsvc+0x01ee (or somewhere
round there). Now start the Server from xdm. We had 680 messages
from one site in a matter of hours, before the system administrator
noticed that the Server was not starting up properly...
Normally, we see problems where the user has misconfigured the Server
(e.g. ignoring our Xsetup tool and manually editing the file,
specifying that you want 1024x768 on a monitor that can only do
640x480). About once a week, we see a problem that we have not
already diagnosed and fixed (we send messages to people that have a
problem that we have fixed, telling them where we have the patch, or
what to alter in config files to avoid the problem).
We try to follow up each of these oddball problems, though many are
irreproducible. In 1.2Beta/2, we could not reproduce a problem
involving a particular Server failure in a font handling function.
We realised the cause when we asked a second user that developed the
problem and from then on, rapidly traced the cause to something in
xmcd. This and a few other instances would have been impossible to
find using normal reporting methods.
> > On the basis of prior experience, we found that the proportion of
> > inexplicable system failures dropped significantly when we switched
> > to using UPS's on all machines. Under the category of "one of those
> > strange things", count very short power supply interruptions. I'm
> > not sure about your power supply company, but in the UK and the US,
> > they don't count interruptions of below a couple of seconds. That is
> > a delay long enough to cause most systems to reboot. If shorter, you
> > may simply get a bad sag. At the other end, could you have received
> > a spike in the voltage - are you near a Physics Lab, for example :-)
>
> Though I'm in a physics lab our power is very stable (despite of some
> announced outages once or twice per year) - power is very constant
> in Europe anyway compared to the US due to the Energie Verbundsystem.
Last time I checked, the UK was still reasonably close to Europe. I
was, at one time, a University Lecturer (equivalent to a Professor,
to you Yankees). I did a study on control systems in the electricity
supply industry. The UK voltage was nominally 240VAC. The Central
Electricity Generating Board normally maintained 250VAC, with
excursions up to 253VAC permitted... Most American hardware companies
produce power supplies that are 110/220VAC. Most power supplies have
about 10% tolerance. 220VAC+10% ~= 242VAC, which would mean that a
true 240VAC supply would be safe. A 250VAC main supply exceeds
tolerances and cause shortened power supply life and often results in
power supply variations being propogated.
Later, working as a troubleshooter for a computer company, I found
that one Banks' computers were crashing overnight, almost every
night. The local electricity company denied any power outage. A
line monitor revealed that every night, between 2am and 3am, there
was a 2 second power supply interruption. This was eventually traced
to an operating procedure at the local electricity provider; the
customer bought a UPS and was much happier. Further investigation
revealed that very short outages, that would not significantly affect
a fridge or fan, are considered normal practice in the electricity
industry, in the UK and the US at least, and are not counted as an
outage.
Then, working in Switzerland, porting an application to a bank of
machines, I had similar unpredictable power supply variations... So
Germany may be ahead of the Swiss in this technology... or the
nearby factory with it's heavy welding equipment my have been
responsible.
So, before you claim a nice power supply, make sure that you don't
have people striking arcs on welding torches, and so on. It's
amazing what they do to a power supply. You might not get an outage,
but you'll get dreadful sag and a lot of spiky noise.
Cheers, JeremyC.
--
Jeremy Chatfield, +1(303)470-5302, FAX:+1(303)470-5513, email:jdc@xinside.com
X Inside Inc, P O Box 10774, Golden, CO 80401-0610, USA.
Commercial X Server - for more information please try these services
http://www.xinside.com info@xinside.com ftp.xinside.com
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199501190904.CAA14730>