[Nut-upsdev] NUT shutdown on 'building alarm' (

Kjell Claesson kjell.claesson at telia.com
Sat Jun 3 23:12:28 UTC 2006


lör 2006-06-03 klockan 13:42 -0500 skrev Bruce Allen:
> Kjell,
> 
> Thanks for your note.  I put my comments in-line below, with a couple of 
> questions.
> 
OK.
> That's exactly what I had in mind.  If I go this route, do I also need to 
> 'clear' some flag if the alarm 'goes away'?
> 

No. It clears when the condition is restored. So if the temperature 
drop down it clears.

But there is a 'alarm history block' that holds information on alarms
and notifications. (Reading of this is not implemented due to the 
amount of data that it need to read.)  

> Our normal strategy is that if our UPS is on battery for more than 90 
> secs, we initiate a shutdown.  We want the same behavior for the 
> thermostat.  So if a 'glitch' causes the thermostat to close for 10 secs 
> then the systems are not shut down.
> 

That should be Ok if you link the alarm to issue OB LB as it clear this 
at next poll of info from the ups, when the condition is cleared.
But You are on your own. I would not do it like this. You can not 
handle this building alarm different than a shutdown condition OB LB.

Good idea from Arjen, to run a generic driver beside to handle external
alarms.

> > I think it would be better to make the upsmon handle alarms. Then the 
> > code is in one place and drivers that support alarms should support 
> > sending them.
> 
> Obviously I would be very happy to see this!  In general it would be very 
> useful.  For example some sites might want to shut down if the power 
> frequency or voltage does outside some fixed limits, or if some UPS 
> components generate temperature alarms from being too hot.
> 

There is info strings declared like 'ambient.temperature.alarm' but it 
is handled like the other info from the ups. And this is not supported
in the driver to set this info. 
So this is something that would be put in the pipe for future 
releases.  


> > The time has been limited for me to work on the driver. It should 
> > support 3-phase info by now and some other fixes, but it's in the pipe.
> 
> Thank you for your work on this driver.  It's been very helpful to us.
> 
Thank you, nice to know. But some of the hard parts was written by Tore
and the usb support is from Wolfgang, and all started when Tore and I
wanted our Powerware ups to run on Linux with NUT.  

> Am I correct that supporting alarms requires 'widespead' changes to NUT, 
> not merely changes to the driver?

No, it would take some more efforts then just the driver.

Regards
Kjell





More information about the Nut-upsdev mailing list