[Nut-upsuser] System with MGE UPS shuts down too early

[Georgi D. Sotirov]

OK, so yesterday evening I done a real test cutting of the power to the 
UPS. And it went good... the UPS supported my server for 28:50 minutes 
(i.e. the expected runtime with this load), before forcing shutdown. The 
batteries could still hold as charge was 30 % with about 15 minutes run 
time. And the UPS did hold for another full 10 minutes before powering 
off. These are the relevant lines from /var/log/ups:

20200124 201318 32 0.0 12 [OB] NA 0.0
20200124 201323 30 0.0 13 [ALARM OB LB] NA 0.0
20200124 201328 30 0.0 13 [FSD ALARM OB LB] NA 0.0
20200124 201333 30 0.0 13 [FSD ALARM OB LB] NA 0.0
20200124 201338 30 0.0 13 [FSD ALARM OB LB] NA 0.0
20200124 201340 30 0.0 13 [FSD ALARM OB LB] NA 0.0

The server was shutdown properly, but for some reason it did not power 
off. I saw an error from umount about busy file system, but I'm sure 
this doesn't always happen.

With everything powered off, I extracted the original batteries to check 
them and measure the voltage. The batteries are Leoch DJW12-9.0 with one 
of them at 12.57 V and the other at 12.64 V (measured without load after 
the discharge).

And this morning there was again a short power failure (not more than 30 
minutes, because the router connected to the battery power from the UPS 
did hold up). With batteries charged up to 91% the UPS supported the 
server for just 07:10 minutes and forced shutdown with 69% battery 
charge and over 20 minutes of run time...

20200125 021123 70 0.0 16 [OB] NA 0.0
20200125 021128 69 0.0 16 [FSD ALARM OB LB] NA 0.0
20200125 021133 69 0.0 16 [FSD ALARM OB LB] NA 0.0
20200125 021138 69 0.0 16 [FSD ALARM OB LB] NA 0.0
20200125 021143 69 0.0 16 [FSD ALARM OB LB] NA 0.0
20200125 021148 69 0.0 16 [FSD ALARM OB LB] NA 0.0
20200125 021150 69 0.0 16 [FSD ALARM OB LB] NA 0.0

So, it seems to me that my UPS forces shutdown pretty randomly. Why the 
UPS is not waiting for the preset low battery charge value of 15%? What 
is actually driving this FSD ALARM and LB signal when batteries for sure 
could hold up more?


Regards,
--
Georgi

On 24.1.2020 at 18:28, Manuel Wolfshant wrote:
> On January 24, 2020 6:13:31 PM GMT+02:00, "Georgi D. Sotirov" <gdsotirov at gmail.com> wrote:
>> OK, my UPS turned out to be older that I thought. I apparently bought
>> and installed it in September 2016 (not 2017 as I initially wrote).
>> That
>> makes for over 3 years of work, which is over the life expectancy of
>> the
>> batteries,
> Good batteries can last 4-5 years. The chinese ones available in most shops last cca 2 years.
> I use Panasonic (ordered from SOS Electronic) and they last 3-4 years. Even more.
>
>> so I ordered new battery pack.
>>
>> However, how could I make a full charge/discharge test? Should I
>> disable
>> SHUTDOWNCMD in upsmon.conf? Or is there any other way to ignore forced
>> shutdown? I'd like to see how long it would go until the batteries
>> actually drain.
>>
> Disconnect the computer from the UPS, connect a dummy load (like 2-3 * 100W tungsten light bulbs) , turn off the nut client on the computer. Start a stopwatch. Unplug the UPS from mains and wait for it to die. Read the value on the stopwatch.
>
> wolfy
>
> _______________________________________________
> Nut-upsuser mailing list
> Nut-upsuser at alioth-lists.debian.net
> https://alioth-lists.debian.net/cgi-bin/mailman/listinfo/nut-upsuser

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://alioth-lists.debian.net/pipermail/nut-upsuser/attachments/20200125/a2815020/attachment.html>