Nut upsuser - Dec 2007 - Strange battery power shortage on an Unitek iZi UPS 525

Hi everyone,

I've been using nut 2.0.3 for more than a year on a Unitek iZi UPS 525
(which is a rebadged Mustek). Everything worked fine and last week, when
pressed by RENEL :-) I ran a battery test which showed the system to have
17min of back-up power (yes, still haven't figured out how to use
upssched...)

The system used for testing was an old PIII/500MHz/512KB cache, 256MB RAM (3
x SDRAM 100MHz) 1 x 8GB hdd IDE and a 250W PS. For various reasons I've
switched to a new system consisting of one Duron 900MHz, 1 piece of 512MB
RAM DDR1 and a 12GB IDE (same family as the previous 8GB). After reinstall,
an upgrade to "nut-2.2.0-6.1.fc8", copying of the same nut config
files and
editing them so that the driver is now megatec instead of powermust I've
discovered that the battery time is down to a mere 4 minutes. Any ideas are
welcome, so far I can only think of:

1. The battery has lost its power in less than a week since my last test

2. Duron 900/64KB cache chews three times more power compared to the old
PIII/500. I strongly doubt this one but I'm willing to hear other
opinions...

3. Somehow the newer version of nut fails to correctly read/interpret the
values from the UPS and decides to shut down earlier.

4. ... any other ideas?

Oh, since I no longer have the external DB9 connector for this motherboard
I've resorted to a USB2Serial from "Sandberg". Apparently nut
doesn't seem
to care if the port is ttyS0 or ttyUSB0...

PS. I've also replaced the case; can this be the reason? :-]

On Dec 22, 2007 2:34 PM, Ciprian Marius Vizitiu <cvizitiu at gbif.org>
wrote:
> 3. Somehow the newer version of nut fails to correctly read/interpret the
> values from the UPS and decides to shut down earlier.
Unless you specify a "lowbatt" value in you ups.conf, the shutdown
will happen when the UPS raises the "low battery" flag. How this flag
is handled by megatec/powermust hasn't changed between 2.0.3 and
2.2.0.
> 4. ... any other ideas?
Either you "new" system draws enough extra power, or your UPS
wasn't
fully charged when you did that second test, or your battery is going
bad, or a combination thereof.

It doesn't look like something related to NUT.

-- 
Carlos Rodrigues

On Dec 22, 2007, at 9:34 AM, Ciprian Marius Vizitiu wrote:
> The system used for testing was an old PIII/500MHz/512KB cache,  
> 256MB RAM (3
> x SDRAM 100MHz) 1 x 8GB hdd IDE and a 250W PS. For various reasons  
> I've
> switched to a new system consisting of one Duron 900MHz, 1 piece of  
> 512MB
> RAM DDR1 and a 12GB IDE (same family as the previous 8GB).
To add to Carlos' comments, I think the combination of higher  
processor speed and higher memory bus speed (SDRAM vs. DDR) could at  
least double the load. If the battery is getting old, and you are  
doing deep discharge testing, that drop in runtime is well within the  
accuracy you can expect from a low-end UPS.