Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 15 Oct 2013 12:09:15 +0100
From:      Frank Leonhardt <frank2@fjl.co.uk>
To:        freebsd-questions@freebsd.org
Subject:   Re: UPS buying suggestion
Message-ID:  <525D225B.7050205@fjl.co.uk>
In-Reply-To: <CACo--msUpY-6r7MkuEvrPDpSVdFZyBotSA-eS7aLGMFDeq_vDQ@mail.gmail.com>
References:  <CACo--msUpY-6r7MkuEvrPDpSVdFZyBotSA-eS7aLGMFDeq_vDQ@mail.gmail.com>

Next in thread | Previous in thread | Raw E-Mail | Index | Archive | Help
On 15/10/2013 06:07, yudi v wrote:
> I am planning on buying an UPS to protect my HP microserver
> n40l<http://n40l.wikia.com/wiki/HP_MicroServer_N40L_Wiki>,
> it will be running FreeBSD 9.2 RELEASE.
>
> I am looking at
> APC Power-Saving Back-UPS ES 8 Outlet 700VA 230V AS
> 3112<http://www.apc.com/products/resource/include/techspec_index.cfm?base_sku=BE700G%2DAZ&total_watts=400>;
>
> is that supported by apcupsd? (already tried apcupsd mailing list, there
> was no response, hence asking here)
>   If not, please recommend one that is supported.
>
> The HP server has a 150W PSU and the UPS is rated at 400Watts, and it comes
> with USB monitoring support.
>
> Also, how to measure exact power consumption by a device?
>
>

I've been watching this thread with a bit of interest. For what it's 
worth I run several UPS units at various location (about forty) and FWIW 
I tend to avoid the cheap ones although I've not actually had a problem 
with them. It's just when I open the case I don't like the manufacturing 
quality. I'm using (right here) an APC BackUPS Pro 420 to keep three HP 
microservers alive. What!? Yes, it works. The power they draw varies 
greatly depending on what type and how many drives you have in the bays 
and (I think with the Microservers) their processing workload.

As to the "compatibility", I do the myself. All I want it to do is shut 
down gracefully if the power fails - nothing more. And this is how I do it:

All servers are, by definition, connected to the network. They can ping 
equipment nearby (just do it from a shell script). If they ping several 
pieces of kit on the LAN that don't have a UPS and get no reply, 
repeatedly for a few minutes) then it can be assumed something's up, and 
they can shut down. As a cross-check they can ping other kit on a UPS 
and if that's still alive it proves the NIC is okay. In the event of a 
suspected failure the script writes what it's done in a log, emails it 
to me (for later) and calls shutdown.

Why don't I use a special driver? First off, one UPS may be powering 
more than one server but only one can be connected (and I've got a few 
*BIG* UPS units covering a whole rack). Secondly, UPS units need 
servicing, which means they get swapped. Unless you have just one type 
you've got more than one protocol to contend with, and is that USB or 
RS-232? It quickly becomes a pain. I want to be able to swap any UPS for 
any other and still have automatic shutdown without messing with the 
server hardware/software.

So, what I look for in a UPS is a good automatic self-test (i.e. a 
"smart" one) and a standard and easy to change battery (the 7Ah type are 
cheap, plentiful and are thus well refined).

As to the "run time", there are lies, damn lies ans statistics. I once 
tested a load of them for an article in PC Magazine and took very little 
notice of what the manufacturer said. They normally have the same 
battery, and without breaking any laws of physics they're going to run 
for pretty much the same time unless they're using a cheap and 
inefficient inverter, in which case the energy gets dissipated as heat 
and you'll know it. Inverters are a done deal now - just buy the chip - 
I wouldn't expect much variation. The only performance difference you're 
likely to find is the output current, and the transistor it flows throug 
- if it's weedy it will overheat/burn out, but that takes time. You pay 
more for sustained higher output currents, but may very well not need 
them. As to the run-time, that's always down to the battery capacity. 
Buy a new one every two years; don't wait for the load test to fail.

Regards, Frank.










Want to link to this message? Use this URL: <http://docs.FreeBSD.org/cgi/mid.cgi?525D225B.7050205>