Re: Debian Compatable UPS?
On Sun, 09 Oct 2005, Alvin Oga wrote:
> > > the answer is all ups will work if you have the magic wand ( skills )
> >
> > This is incomplete. You also need magic powder (time and other resources),
> > which are often not of trivial cost. So this answer is at best useless.
>
> like all things ...
> - some folks can do it in 5 min ... some takes 5 hrs
Don't give me that. I *know* how to do it, and the correct answer is: "for
some units it takes 5 minutes because they are just silly things that you
just need to folow the traces with a multimeter to find out their function".
But for a more complex microcontroller-based unit, it can take several days
(and I don't mean just a couple of them) to reverse engineer the full
smart-signaling protocol.
The magic powder can be *very* expensive, even if you already are a wizard
with a good magic wand.
> - knowing which screws to turn will costs $100/hr as the saying
> goes (the home car mechanic vs certified(nutcases) at the dealers)
I am a certified(nutcase) when it comes to UPSes by your definition, I
suppose. I am an electrical engineer (fully certified) after all...
> > His claim is right, batteries will last at most three years (do not expect
> > to get one that will last that much with any consumer or prosumer grade
> > UPS). UPS units can, and often last up to ten years (but not consumer grade
> > ones).
>
> some of my el cheapo consumer stuff is still working after 8hrs ...
I will suppose you meant 8 years, not 8 hours.
> - been tested about a month ago with an unexpected 5minute power
> outage
Lucky you. You have good batteries on that one, write down their model and
manufacturer to get more of the same kind later.
> - if the machines is shutdown by the ups,
> - who/what is gonna go turn it back on later ..
If you have a decent UPS, *it* will turn the machine back on later, after
main power returns and it has charged the batteries enough.
> - how do you know its safe to come back online
Because you got a proper UPS and programmed it to not turn the load on
before its batteries are at least 30% full, I suppose.
Sure, we are assuming the UPS is not breaking down and doing dangerous silly
things. I have never dealt with one who would screw up on this, though,
even the cheap ones would not turn the load on line power while it was out
of range.
> - did the machine have enough time to shutdown properly or did
> it die cause it waited too long
Depeds on whether people are doing stupid things on the shutdown scripts,
etc. Test it, any proper UPS monitor package can be ordered to simulate a
low battery powerdown to test the software end of things. Do make sure to
test it with the network down as well.
> > An UPS can easily destroy the load in certan failure modes, and it does not
>
> those are the socalled "bad ups"
It could be a good UPS breaking down, as well... but this should indeed be
rare.
> some ups claim to be able to protect your equipment against lightening
> strikes ...
> - the folks in the mid-west and mid-atlantic gets to test
> their PCs and ups on a regular basis
Against indirect lightning-induced surges? Sure. Against *direct* ones? I
doubt it very much so, unless we are talking about US$500+ UPSes (and
probably even then...). Proper lightning protection for small stuff costs
at least US$100, and it requires a very good ground to actually work (which
must be far away from the lightning arrest array's grounding or it won't
work).
So, this is not something you should expect to find even on prosumer models.
> > but it CAN provide enough surge suppresion to
> > protect the load and its control channel from surges induced by the lightning
> > EM-field in the internal wiring.
>
> for surge protection ... a good $10 surge protector works just as good
$10 surge protectors, *IF of very good procedence*, will get the induced or
residual surges that get past *proper* lightning surge protection. These
small surges your UPS should already be capable of handling. And your PSU
should too, if it is not crap.
I would NEVER recommend anyone to buy a US$ 10 surge protector and believe
they are even remotely safe from lightning damage because of that US$10
gadget...
If you have no real lightning protection (e.g. you live in a house), and
want it, you need a valid configuration made of stuff like this:
http://www.citelprotection.com/citel/AC_EL.htm
Note: it *will* cost a lot more than US$ 10, try US$ 150 or so, maybe more.
And you need very good grounding too, which is expensive as well and
requires maintenance.
> and for good high end ups .. powerware/lieberts are better
Agreed. The big Powerware units are the best you can get for datacenter-
grade UPS around here, AFAIK. And the capacitive coupling for redundant
units is very nice, not to mention the fact that they have proper battery
control (I don't buy into their marketing that they are the only ones to do
so, though :-P ).
You *do* pay for the quality though. I had a quote of about US$ 25k for a
60kVA unit about an year ago (here in Brazil).
> in 90% of the cases, i'd avoid the $200 - $1500 ups market
> if it's my pesonal nickel to spend
> - for $1000 ... i can have 2-3 identical complete systems
> ( $80 mb, $80 cpu, $40 disks, ... to protect data )
<shrug> That wouldn't work for me. If you need it online while you're out,
you'd have all three units fried in an year without an UPS, PLUS you'd get
downtime and constant crashes which are certainly annoying even if you don't
lose any data permanently.
--
"One disk to rule them all, One disk to find them. One disk to bring
them all and in the darkness grind them. In the Land of Redmond
where the shadows lie." -- The Silicon Valley Tarot
Henrique Holschuh
Reply to: