Air Conditioner casuing reboots and blue screen errors? Is this possible...
Pej
One of our web servers keeps on rebooting and getting different stop
errors each time. 0x1e, 0x24, 0x0a, etc. Ive been working on this for
months on what is causing this. Ive tried one by one replacing hardware,
RAM, different RAM slots, but whatever Ive tried just dont know what is
causing the stop errors.
Right above the server is a air conditioner to keep the server room cool,
(The other servers a seperated by a wall). Last week the air condition
broke and is not in operation now until repair people come next week to have
a look at it. This broke and stopped operating about 3 days ago and the
server has not rebooted once. Before it would reboot at least 1 - 2 times a
day and sometimes 5 - 6.
It sounds crazy but is it possible the air conditioner was causing these
stop errors and reboots? Maybe somehow it was causing interferance to the
hardware??
Thanks in advance
Jason W
news:1vkb8.30207$Zu6.1...@news-server.bigpond.net.au...
I am definitely no expert on this, but I don't see why not. AC causing noise
when it cuts in and out could definitely be interfering if one or more
components in the system is already near their tolerance levels. But again,
this is just my opinion and I have no technical knowledge on the subject.
Jason
KlausF
- The air conditioner causes voltage dips when it cuts
in. Get a small UPS for your server - they don't cost much.
- The air conditioner causes electrical interference, either
through the mains or through the air. If it comes through
the mains then a UPS will probably fix it. If it comes through
the air then you need to relocate your server. You must
also make sure that it is properly earthed - consult an
electrician.
- Your server does not like the low temperatures produced by
the air conditioner. Increase the temperature setting, and
don't expose the server to direct blasts of arctic air.
- You get condensation because of the low air temperature
coming out of the air conditioner. Don't expose the server
to direct blasts of cold air.
- Your air conditioner is dripping water onto the server.
Good luck!
"Pej" <nag...@bigpond.net.au> wrote in message
news:1vkb8.30207$Zu6.1...@news-server.bigpond.net.au...
Andrew Rossmann
copy was sent to the cited author.]
In article <1vkb8.30207$Zu6.1...@news-server.bigpond.net.au>,
nag...@bigpond.net.au says...
It almost certainly is. The most likely reason is the sharp voltage drop
and surge when the compressor kicks on and off. The easist fix for this is
a good UPS (Uninteruptible Power Supply.) Get one with line voltage
regulation, not a cheap, basic, battery only model.
I certainly hope the A/C and computers are not on the same circuit! If
so, try to separate them. Make certain everything is grounded properly.
Try to keep any computer-related cables away from the A/C unit.
Also, as others have mentioned, don't let the cold air blow directly on
the computer.
--
If there is a no_junk in my address, please REMOVE it before replying!
All junk mail senders will be prosecuted to the fullest extent of the
law!!
http://home.att.net/~andyross
J. Clarke
So...@TooMuch.Junk.net says...
> You've got several possible causes:
>
> - The air conditioner causes voltage dips when it cuts
> in. Get a small UPS for your server - they don't cost much.
>
> - The air conditioner causes electrical interference, either
> through the mains or through the air. If it comes through
> the mains then a UPS will probably fix it. If it comes through
> the air then you need to relocate your server. You must
> also make sure that it is properly earthed - consult an
> electrician.
>
> - Your server does not like the low temperatures produced by
> the air conditioner. Increase the temperature setting, and
> don't expose the server to direct blasts of arctic air.
>
> - You get condensation because of the low air temperature
> coming out of the air conditioner. Don't expose the server
> to direct blasts of cold air.
>
> - Your air conditioner is dripping water onto the server.
Or there is a loose component in the server that is being disturbed by
vibration from the air conditioner.
>
> Good luck!
>
> "Pej" <nag...@bigpond.net.au> wrote in message
> news:1vkb8.30207$Zu6.1...@news-server.bigpond.net.au...
> > Hey,
> >
> > One of our web servers keeps on rebooting and getting different stop
> > errors each time. 0x1e, 0x24, 0x0a, etc. Ive been working on this for
> > months on what is causing this. Ive tried one by one replacing hardware,
> > RAM, different RAM slots, but whatever Ive tried just dont know what is
> > causing the stop errors.
> > Right above the server is a air conditioner to keep the server room
> cool,
> > (The other servers a seperated by a wall). Last week the air condition
> > broke and is not in operation now until repair people come next week to
> have
> > a look at it. This broke and stopped operating about 3 days ago and the
> > server has not rebooted once. Before it would reboot at least 1 - 2 times
> a
> > day and sometimes 5 - 6.
> > It sounds crazy but is it possible the air conditioner was causing these
> > stop errors and reboots? Maybe somehow it was causing interferance to the
> > hardware??
> >
> > Thanks in advance
> >
> >
>
>
>
--
--
--John
Reply to jclarke at ae tee tee global dot net
(used to be jclarke at eye bee em dot net)
w_tom
incandescent lights were also dimming to well under 50% (36% according to
calculations). No appliance must EVER create such a voltage drop. Such a
voltage drop means the air conditioner is about to create a fire or burn out, OR
the building has developed a serious, dangerous wiring problem, such as loose
wire nuts, that also may create a fire. Clearly a UPS is not a solution to
either.
If an incandescent bulb is not dimming to that extreme, then that is not your
problem. Furthermore, no standard appliance must create transients that can
crash a computer. It this is happening, two possibilities. First is a defect
in the air conditioner that requires replacement. Second is a poor grounding
system, maybe aggravated by the first point, that should never have been wired
that way in the first place. This second point will be expanded on because of
its complexity.
An old story in DataMation magazine of a tech who ran diagnostics constantly
on a computer without failure. But the computer failed repeatedly in customer
use. Only when he left for coffee did the computer crash on diagnostics. It
got to the point that he would leave the room, loudly proclaiming that he was
going for coffee, then sneak back to peek around the door - to see if the
computer failed. It never did.
The computer was grounded improperly to share with only one elevator. Only
when the tech used the elevator for coffee, then the computer crashed.
Digital ICs are little effected by + side power variations. But they are
sensitive to ground plane variations. A server motherboard connects direct to
building grounds. If this safety ground is improperly connected, as especially
if the air conditioner has developed unacceptable ground leakages, then you can
have computer problem. Of course those ground leakages can only create failure
if there is another problem.
The UPS recommendation is flawed. Although, in theory, a low voltage can
cause a crash, the reality is that no such appliance generated, low voltage must
ever exist. If the air conditioner were suffering sub 90 VAC voltages, then
the air conditioner would not startup. For example, how often does any
starting motor dim lights to well below 50% intensity. Your would be demanding
your electrician fix the wiring. That would be a human safety and a fire
safety problem that the UPS would not fix. The UPS is recommended is flawed -
only to repair the symptom - not the problem.
Also that UPS cannot make any changes in a bad safety ground connection.
Ground from motherboard, through UPS to safety ground, must be solid - no
filters or impediments to that connection. That is standard electrical
knowledge. If the outlet ground is somehow improperly connected, then the UPS
makes no changes to the reason for failure.
A last month editorial in an engineering magazine noted how most engineers
that specialize in this problem are bald. Not because this type of electrical
problem causes loss of hair - although it should. It is because such problems,
requiring trivial solutions, require so much sweat, knowledge and experience to
permanently eliminate. Appreciate why this problem can be so complex to find -
and the solution be so trivial. Best to start with basics. How does the air
conditioner and servers connect to break box panel. How are their ground wires
and neutral wires routed to connect to the breaker box. What is different in
the power circuit to one verse power to all others. Does the one circuit share
a neutral wire with a completely different circuit to the air conditioner? What
changed in this air conditioner that it is now creating more electrical noise?
Remember a server can startup and run even when the 120 VAC is well below at 90
VAC (assuming the server power supply is not suffering a serious part
degradation). No motor will even create too low voltage for a server - except
where theory is not tempered by reality.
Step one is the eliminate the reason why an electrical noisy air conditioner
can cause failure. That server should be so resilient that all noisy air
conditioners can never create failure. Makes this and all future servers more
resilient. Then replace the air conditioner or server power supply. Neither
must permit the failure even if the other problems existed. (BTW, those sub
$100 power supplies have so many electrical functions missing that failure among
clone computer builders is situation normal.) Something has changed. To
actually say which changed requires someone who routinely solves problems with
tools like oscilloscopes. Most computer techs don't even have O'scopes let
alone know how to use them. Therefore they would solve this problem with wild,
shotgun solutions such as dedicated power circuits or more fans, or UPSes. You
must solve the problem - not the symptoms. As noted, one common reason for such
failures are power supplies that only a bean counter could love - so much less
expensive because basic functions are missing.
BTW, another has also mentioned a valid alternative - air conditioner
vibrations increased to finally cause that server's manufacturing problem to
appear. But forget the UPS suggestion. It is based on theory made totally
irrelevant when numbers are applied. At best, the UPS can only solve a symptom
to a much larger problem. Check for obvious such as compromised grounds,
failing air conditioner that shares a common neutral with only that server
(recently air conditioner created harmonics that should never cause a computer
problem if the power supply was properly designed), or plug an incandescent bulb
in the server's outlet to see if it dims to 50% when the air conditioner turns
on - a serious air conditioner or building wiring failure.
Furthermore, what is normal operating temperatures for any computer? 40 to
100 degrees F. Room temperature and variation will never cause properly
manufactured computers to fail. Failure do to operate with room air temperature
changes means a computer failure - one that can be proven by running the
computer in a 100 degree room and/or in a 40 degree refrigerator. Yes, we did
just that type of experimenting when our products could never fail or suffered
intermittent failure. But today, it should not be a problem and can easily be
proven by the above simple experiments.
KlausF
Air conditioners use squirrel cage motors. Such motors
have an inrush current at startup of up to 6 x nominal current.
This can easily cause a substantial voltage drop. Since
the drop lasts for only half a second there is no risk of overheating.
A UPS will bridge such a dip with ease.
And no, good or bad grounding has nothing do to with it.
Grounding is for safety; it does not carry load currents.
"w_tom" <w_t...@usa.net> wrote in message news:3C6FAF91...@usa.net...
|8ish|
equipment
sharing the ground, especially if the ground is not sufficiently "grounded".
"KlausF" <So...@TooMuch.Junk.net> wrote in message
news:a4pakn$du0$1...@arachne.labyrinth.net.au...
w_tom
wiring is insufficient for the load. No motor should cause incandescent lamps
to dim to 36% intensity for 1/2 second. If KlausF was correct, then dimming
lights would be normal everyday.
No motor can cause so much voltage drop for that tremendously long 1/2
second. That excessively deep voltage drop for an excessively long 1/2 second
could even be symptoms of a human safety problem. If motors are dimming lights
below 50%, then call the electrician immediately. If motors are dimming the
lights for that tremendously long 1/2 second, then turn off that motor before
calling the electrician.
If grounding could not be a problem, then why did that elevator cause computer
crashes? The primary function of grounding is human safety. However there are
currents on all ground line. Disconnect ground from a refrigerator and measure
the refrigerator voltage relative to the disconnected safety ground. If KlausF
was reporting accurately, then those 10 and 100 microamp currents would not
exist, that refrigerator voltage could not be measured, and the elevator would
not have crashed the computer. Clean building grounds are important to reliable
computer operation.
The UPS recommendation is a Band Aid - to cure symptoms of a problem. Since
KlausF 'feels' it is normal for lights to dim to below 50% intensity every time
the refrigerator turns on, well, then there is this minor issue of his
credibility. His UPS recommendation is clearly flawed. Then the urban myth
that cold from the air conditioner will cause a computer crash - despite the
temperature specifications for that computer. As for all this condensation -
where is that coming from when the air conditioner takes moisture out of the air
leaving a hot computer dryer.
KlausF
> If grounding could not be a problem, then why did that elevator cause
computer
> crashes? The primary function of grounding is human safety. However there
are
> currents on all ground line. Disconnect ground from a refrigerator and
measure
> the refrigerator voltage relative to the disconnected safety ground. If
KlausF
> was reporting accurately, then those 10 and 100 microamp currents would
not
> exist, that refrigerator voltage could not be measured, and the elevator
would
> not have crashed the computer. Clean building grounds are important to
reliable
> computer operation.
You miss the point. Grounding has two effects:
a) human safety
b) it tends to keep out interference (which I said all along)
It does not and must not contribute to voltage dips. If it did then
the installation would be unsafe.
> The UPS recommendation is a Band Aid - to cure symptoms of a problem.
Since
> KlausF 'feels' it is normal for lights to dim to below 50% intensity every
time
> the refrigerator turns on, well, then there is this minor issue of his
> credibility.
Thanks for putting words into my mouth. I never said "50%" - this is your
figure.
I said that UPSs are exellent for bridging dips caused by A/C inrush
currents.
> As for all this condensation -
> where is that coming from when the air conditioner takes moisture out of
the air
> leaving a hot computer dryer.
Get your physics straight. When you blow very cold air against a
surface, it will cool down. When you stop the cold blast, room air
flowing past this surface will get cooled down, causing condensation.
You can observe this phenomenon with any cold glass of beer.
