Quick quiz about redundant power supplies
Jurjen Oskam
A) It causes the 8A circuit breaker on that powercord to trip
B) It causes the big-ass 32A circuit breaker to the rack to trip
C) It causes the UPS to temporarily shut down its battery backup due
to overload
D) It causes the entire server to spectacularly fail even though the
other power supplies are on an entirely different power group
with its own circuit breaker
E) All of the above
<sigh> This reminds me of the phrase: "A $100 component will protect a
$0.10 fuse by blowing first."
--
Jurjen Oskam
PGP Key available at http://www.stupendous.org/
Dan Birchall
> What happens when a redundant power supply fails?
>
> A) It causes the 8A circuit breaker on that powercord to trip
> B) It causes the big-ass 32A circuit breaker to the rack to trip
> C) It causes the UPS to temporarily shut down its battery backup due
> to overload
> D) It causes the entire server to spectacularly fail even though the
> other power supplies are on an entirely different power group
> with its own circuit breaker
> E) All of the above
Wait a second... where's the option for the entire building (7 floors
of parking deck, plus ten floors of office space) losing power? None
of the options above mesh with my experience... ;)
Mike Andrews
I saw a location crew from $CBS-AFFILIATE here do pretty much that,
in such a way as to get not just publicity, but also legislative and
judicial notice. _Un_welcome legislative and judicial notice.
The crew were setting up lights for a shoot at the Oklahoma State
Capitol, and were looking for breakers to determine if they were rated
to sustain the draw of the big quartz-halogen lamps. To do this, the
crew chief used a "breaker-finder": a big-ass 3-conductor plug with
all three terminals connected by about 6-or-8-gauge wire. Procedure
was to plug it into the wall, and when someone came out to complain,
have that person guide the crew chief to the breaker box.
But the branch breaker failed to trip on overcurrent, and so the next
line of defense, the _floor_ breaker, tripped, taking everything on
that floor of the Capitol with it. All the peecees, desk lamps, radios,
recorders, overhead lights, and everything else, for dozens of state
congresscritters, judges, and other elected and appointed officials,
as well as a great number of state employees.
Restoring the floor breaker to the "on" postion merely powered up the
line to the breaker-finder, which caused it to trip out again, because
it was by now welded into the socket.
So the contract electricians had to be called out to replace the wall
socket, at Considerable Expense to $CBS-AFFILIATE, and with quite a
long delay.
IIRC, $CBS-AFFILIATE uses a different method of finding breakers now.
--
Bill Gates says Open Source is riding on the coat-tails of Windows' success.
That's like saying Osama has done his part to increase airport security.
-- Carl Schelin, in the Monastery
Graham Reed
> What happens when a redundant power supply fails?
[list deleted]
Damnit, fix your bloody power supplies. You're causing failures at my
house too.
Gives me a great excuse to oversleep; the clock radio isn't on a UPS
(and is the kind with good sound instead of a built-in battery
backup).
--
"Your gutter is on the opposite side of the street from my gutter."
-- adb
Carlo
if they are going to fail they do so with class and style. Why get
someone out to fix it when only one thing is wrong, its more efficient
that the whole machine dies and burns away..
DrCarlo
Jurjen Oskam <jos...@quadpro.stupendous.org> wrote in message news:<slrnb5nfq8...@calvin.quadpro.stupendous.org>...
John R. Owens
I've got a somewhat similar toy: once while exploring an abandoned
drive-in movie theater in the Missouri wilds, I came across an
interesting-looking light bulb in a box. Brought it out with me, and
started experimenting with it. I guess it might have been a projector
bulb or something, although I would not have expected such to fit into a
standard light bulb socket, as this did. And when you do, not only does
it trip circuit breakers/blow fuses, it actually melts parts of the socket
down. >:) I've never yet gotten a chance to see how bright it is lit on a
continuous basis, though.
--
John R. Owens http://www.ghiapet.homeip.net/
Marcus, this is the kind of conversation that can only end in a gunshot.
--Dr. Stephen Franklin
Bron Gondwana
Mike Andrews <mi...@mikea.ath.cx> wrote:
> Dan Birchall <zap.fry...@i.charge.100.bucks.per.email.spam.d1h.com> wrote:
> in such a way as to get not just publicity, but also legislative and
> judicial notice. _Un_welcome legislative and judicial notice.
>
> The crew were setting up lights for a shoot at the Oklahoma State
> Capitol, and were looking for breakers to determine if they were rated
> to sustain the draw of the big quartz-halogen lamps. To do this, the
> crew chief used a "breaker-finder": a big-ass 3-conductor plug with
> all three terminals connected by about 6-or-8-gauge wire. Procedure
> was to plug it into the wall, and when someone came out to complain,
> have that person guide the crew chief to the breaker box.
>
> But the branch breaker failed to trip on overcurrent, and so the next
> line of defense, the _floor_ breaker, tripped, taking everything on
> that floor of the Capitol with it. All the peecees, desk lamps, radios,
> recorders, overhead lights, and everything else, for dozens of state
> congresscritters, judges, and other elected and appointed officials,
> as well as a great number of state employees.
>
> Restoring the floor breaker to the "on" postion merely powered up the
> line to the breaker-finder, which caused it to trip out again, because
> it was by now welded into the socket.
Is it just me, or is this the point where you find the branch breaker
involved and turn _it_ off, then restore power to the rest of the floor?
I'm willing to accept conslutancy fees for my brilliant advice.
Bron.
Mike Andrews
It _would_ have been more intelligent to have found the branch breaker
and turned it off _before_ lighting up the floor breaker again, yes.
It also would have been smarter to pull the breaker-finder before
they tried lighting up any breakers. But we're dealing with a TV
station's location crew, here. Intelligence is *not* a factor in their
activities, and it probably is contraindicated as a criterion in
hiring.
Admittedly, pulling the breaker-finder would have required removing
and replacing the socket, but that's fine. The circuit's down anyway.
--
Downsizing saves money like computers reduce paperwork.
Bron Gondwana
Mike Andrews <mi...@mikea.ath.cx> wrote:
> It _would_ have been more intelligent to have found the branch breaker
> and turned it off _before_ lighting up the floor breaker again, yes.
> It also would have been smarter to pull the breaker-finder before
> they tried lighting up any breakers. But we're dealing with a TV
> station's location crew, here. Intelligence is *not* a factor in their
> activities, and it probably is contraindicated as a criterion in
> hiring.
Or at least as a criterion for remaining for any period of time, which I
guess adds up to the same thing if the hiring people have any sense -
someone who will grow a spine and leave is not a good thing to hire,
from that side of the fence.
> Admittedly, pulling the breaker-finder would have required removing
> and replacing the socket, but that's fine. The circuit's down anyway.
It's also probably illegal unless you're a licenced electrician, at
least in the parts of .au where I am. Given the above TV crew, I doubt
any of them are. Much more sensible (from their point of view) to point
at the Standard Operating Procedure which told them to plug the breaker
tester in, and keep their heads down while the shit hits the author of
said procedure.
Hey, it's what I would do.
The real party at fault here is the manufacturer and/or the installer of
the building circuitry who installed a branch breaker which didn't open
before the floor breaker. Now _that_ is negligence or bad design.
Bron ( and if that's UI to anyone they can go fuck themselves )
AndyC the WB
Jurjen> What happens when a redundant power supply fails?
Jurjen> A) It causes the 8A circuit breaker on that powercord to
Jurjen> trip
Jurjen> B) It causes the big-ass 32A circuit breaker to the rack
Jurjen> to trip
Jurjen> C) It causes the UPS to temporarily shut down its battery
Jurjen> backup due to overload
Jurjen> D) It causes the entire server to spectacularly fail even
Jurjen> though the other power supplies are on an entirely
Jurjen> different power group with its own circuit breaker
Jurjen> E) All of the above
One would hope the answer is F) no-one notices until the log
monitoring system raises the alarm. From the fact that you posted, my
vote is for E) All of the above.
The UPS powering the CSU/DSU for all 4 E1s into this building failed
in the wee small hours of Monday morning. We lost our phone lines,
internet access, and WAN connection. It was tempting to lock the data
center door and act dumb for a couple of hours.
I'd have probably been able to get away with it, too. nPHB is
vacationing in .ca.us, instead of being here in .uk, but would have
been sufficiently clueful to work it out if he'd been here.
AndyC
--
AndyC the WB
Replace SPAM with Andy to reply
Mike Andrews
It's a violation of law in .okc.ok.us to work on electrical stuff
unless you own the structure and/or are a licensed electrician.
I think that the maintenance folks at the Capitol building would
not accept my argument that since the citizens of Oklahoma own the
Capitol, I am entitled to work on the wiring. I'm _sure_ that Shawver
Electric Contractors, who hold the contract for electrical maintenance
and repair, would not.
As to fault: that's a good question. I know that breakers "weaken"
with time and repeated overcurrent trips, so that they tend to trip at
progressively lower current. But I've had no experience with a breaker
*failing* to trip out at all when it was exposed to massive overload.
I'd want to check out the breaker's history in the building, with
special reference to how frequently it had to be reset, and I'd also
want to check out the manufacturer's experience with the breakers in
that part of the product line.
But I'd much rather just fix the fscking problems:
1) Don't induce overloads so that you can find the breaker;
2) People who induce overloads so that they can find the breakers
are *NOT* permitted to use the facilities from that point on; and
3) Replace known-bad breakers.
--
Spamming because of serious financial problems is like
yodeling because your cat destroys your furniture.
-- Gary S. Callison, in nanae
Robert Uhl <ruhl@4dv.net>
>
> As to fault: that's a good question. I know that breakers "weaken"
> with time and repeated overcurrent trips, so that they tend to trip at
> progressively lower current. But I've had no experience with a
> breaker *failing* to trip out at all when it was exposed to massive
> overload.
I would think that if possible breakers would be engineered to fail
safe--that is, rather than failing to trip when they should they'd trip
when they shouldn't, thereby preventing electrical fires, damage to
equipment &c. Dunno if a breaker can be designed in such a way, though.
> But I'd much rather just fix the fscking problems:
>
> 1) Don't induce overloads so that you can find the breaker;
> 2) People who induce overloads so that they can find the breakers
> are *NOT* permitted to use the facilities from that point on; and
I find it rather remarkable that anyone ever put up with such a
procedure--it seems highly annoying and rather dangerous to me. One
would have thought that the first site-owner to experienc eit would have
complained.
--
Consistently separating words by spaces became a general custom about
the tenth century, and lasted until about 1957, when FORTRAN abandoned
the practise. --Sun FORTRAN Reference Manual
Alexander Schreiber
>
>The UPS powering the CSU/DSU for all 4 E1s into this building failed
>in the wee small hours of Monday morning. We lost our phone lines,
>internet access, and WAN connection. It was tempting to lock the data
>center door and act dumb for a couple of hours.
Thats one key advantage of converting your internal phone system to
VoIP: if the notwork ist b0rken, nobody will be able to annoy you
with calls that the network doesn't.
Regards,
Alex.
--
"Opportunity is missed by most people because it is dressed in overalls and
looks like work." -- Thomas A. Edison
Tanuki the Raccoon-dog
<a...@usenet.thangorodrim.de> said
>AndyC the WB <sp...@cunningham.me.uk> wrote:
>>
>>The UPS powering the CSU/DSU for all 4 E1s into this building failed
>>in the wee small hours of Monday morning. We lost our phone lines,
>>internet access, and WAN connection. It was tempting to lock the data
>>center door and act dumb for a couple of hours.
>
>Thats one key advantage of converting your internal phone system to
>VoIP: if the notwork ist b0rken, nobody will be able to annoy you
>with calls that the network doesn't.
The other advantage comes when you configure your IP-phone to
be at 127.0.0.1
--
++Tanuki++
"Now Available: Alex Chiu's new book 'Ten Steps to Surviving the Heat-
Death of the Universe'. Lifetime, money-back guarantee free to the
first 100 purchasers" --from _A_W_O_L_
Michael Hudson
> But I'd much rather just fix the fscking problems:
> 1) Don't induce overloads so that you can find the breaker;
> 2) People who induce overloads so that they can find the breakers
> are *NOT* permitted to use the facilities from that point on; and
s/use the facilities/breathe/
?
M.
--
ARTHUR: But which is probably incapable of drinking the coffee.
-- The Hitch-Hikers Guide to the Galaxy, Episode 6
AndyC the WB
Alexander> Thats one key advantage of converting your internal
Alexander> phone system to VoIP: if the notwork ist b0rken, nobody
Alexander> will be able to annoy you with calls that the network
Alexander> doesn't.
Actually, another approach would be for me to "trial VoIP technology".
Everyone else can stay on the traditional PBX, and I can blame
... er .. high packet latency in the Layer 4 switches when no-one can
call me.
This is of course, for values of "trial VoIP" which mean "putting a
Cisco IP phone on my desk with some letters and numbers inked in on
the LCD display, but not connected to anything".
Richard Bos
> But I'd much rather just fix the fscking problems:
> 1) Don't induce overloads so that you can find the breaker;
> 2) People who induce overloads so that they can find the breakers
> are *NOT* permitted to use the facilities from that point on; and
Indeed; but I'd change this to "People who induce overloads so that they
can find the breakers will be used to find the next breaker". One cannot
overestimate the value of true poetic justice.
Richard