[Nut-upsuser] UPS reboots with APC UPS

Roger Price roger at rogerprice.org
Sat Sep 5 09:15:29 BST 2020

On Thu, 3 Sep 2020, stefanos at karandreas.gr wrote:

>> I see, I wasn't aware of that part.
>> While that holds true for the 1st example-case I stated on the 2nd one 
>> it was certainly couple of minutes before the ups did it's click sound 
>> (thought it was reboot but as you corrected me it's cutting power to the 
>> plugs). That could have been a fluke though?

Curious!  Worth another test to see if it repeats.

>> M2 is usually suspended when I am not physically there. Power cuts out. 
>> After x minutes LB is reached and M1 is set to suspend while power is 
>> not cut of the plugs of the UPS unit at any point in time. UPS operates 
>> with minimal load on it (basically two monitors from M2 that are on 
>> standby). Power comes back, everything continues to be as is. The resume 
>> from suspend will be manual (i.e. click the power button). I am fine 
>> with this approach and should work just fine right?
>> There is also wake-on-lan that works on suspend but would require a 
>> raspberry-pi or something similar.

NUT works as a result of a cooperation between the logic in the UPS, the logic 
in the BIOS and the logic of NUT.  But the whole business of suspension and 
hibernation is managed internally by the BIOS, to which NUT does not have 
access, and there is no possibility of cooperation with NUT.

>> As I think more of it, with the information you provided, the NUT 
>> approach is better as it allows for unattended restart of the machines 
>> regardless of downtime (well unless it exceeds the battery life). 
>> Regrettably the fact that M2 is suspended when I don't use it, is 
>> incompatible with that approach and not using suspend on M2 would mean 
>> starting up everything on M2 at least once per day.
>> Do you know whether ups.delay.shutdown -1 or ups.timer.reboot -1 
>> neutralize it? Or I will have to trial & error?

This is very probably trial and error.


More information about the Nut-upsuser mailing list