Nut upsuser - Jul 2007 - abusing nut ups to do temperature shutdown WHILE monitoring UPS'es

Hi

I am thinking about using the existing NUT software to tell the slave  
servers
to shutdown if the temperature in the server room becomes too high.

I was thinking of monitoring the temperature on the master NUT server  
and
then just letting the slave clients shut down. But the slaves (and  
master
server) needs to do a complete power off even though the UPS's does
get 220v power all the time. Does nut do a complete power off by default
for slave clients?


The reason i want to do this is because what if the master UPS server  
gets
too hot, and shuts down. But that means all the nut slave clients  
looses connection
to the ups master server, and then later the power to the UPS's are  
lost.
Then noone will tell the NUT slave clients to shut down gracefully.


I am open to other suggestions if anyone has (better) ideas.




JonB

> I am thinking about using the existing NUT software to tell the slave
> servers to shutdown if the temperature in the server room becomes too
> high.
If your UPS has a temperature sensor, that's possible. You could script
something that polls 'ups.temperature' every minute and call 'upsmon
-c
fsd' once a certain threshold is reached. Note however, that in most
cases, the power will be cycled then and the systems connected will be
rebooted. So unless you've configured them to stayoff when the power is
restored, this will keep on repeating, unless you send a command to the
UPS to power it down permanently. The latter may be tricky (read on).
> I was thinking of monitoring the temperature on the master NUT server
> and then just letting the slave clients shut down. But the slaves (and
> master server) needs to do a complete power off even though the UPS's
does
> get 220v power all the time. Does nut do a complete power off by default
> for slave clients?
Yes. However, the 'upsdrvctl shutdown' command will in many (if not all)
cases just cycle the mains if it is not actually lost. So you'll need to
find a way to keep it off when the problem was over temperature. This may
be possible, but there are many pitfalls. The 'shutdown.stayoff' command
is not supported by all drivers and if it is, may power off the UPS
instantly. If your server happens to receive power from that same UPS, the
latter may not be desireable. Note that you can only send instcmd's to the
UPS when both driver and server are running, so you may need to fire these
up again in your halt script.
> The reason i want to do this is because what if the master UPS server
> gets too hot, and shuts down. But that means all the nut slave clients
> looses connection to the ups master server, and then later the power to
> the UPS's are lost.
The problem here is that once you've switched off all your equipment, it
will stayoff until someone intervenes.
> Then noone will tell the NUT slave clients to shut down gracefully.
>
>
> I am open to other suggestions if anyone has (better) ideas.
I think a better idea would be to monitor the temperature in the server
room through other means and raise an alarm / send notifications to an
operator who can then decide what to do.

Best regards, Arjen
-- 
Eindhoven - The Netherlands
Key fingerprint - 66 4E 03 2C 9D B5 CB 9B  7A FE 7E C1 EE 88 BC 57