[Nut-upsuser] Maybe this will make it to the FAQ :-)

[Aleksandr Derevianko]

Charles Lepple wrote:

>On 6/24/07, Ciprian Marius Vizitiu <cvizitiu at gbif.org> wrote:
>  
>
>>Chances are the machine is busy running scripts
>>before nut gets a chance to start and anyway the second power failure
>>strikes with no juice in the batteries.
>>    
>>
>
>Some UPSes provide a charge threshold so that the output of the UPS
>does not turn on until after the batteries have charged to beyond,
>say, 30%. If that threshold is adjustable (as I believe it is on
>several MGE models), you can be reasonably sure that the UPS will have
>enough runtime to allow the machine to boot to the point where NUT can
>detect a future power failure.
>
>Your scenario is definitely something to consider if the hardware does
>not provide this option, however.
>
>  
>
And sometimes this build-in 'feature' can hirt. For example, we have a 
industrial environment, and our equipment designed to
survive unclear shutdown and boot very fast - i.e. 10 seconds.
We use UPS'es to workaround downtime in case of short failure, to make 
the power input better and to isolate
hardware from the problems in the field. In Eastern Siberia it is 
frequent problems with not-very-good power lines.

As result, in this environment we need to configure UPS'es to stay on 
battery as much as possible, and set minimal threshold (<5%)
at power restore.

Indeed, some MGE models have a bad idea to decide that nobody is sucking 
it's power if current is small. For example,
we once have a 1600VAC UPS to power only one switch (other equipment was 
broken at the moment and switched off).
If power fail, this UPS turns off output very fast.

-- 
С уважением,
Александр Деревянко