On Nov 29, 2021, at 11:58 AM, Ken Brubacher wrote:
>
> Is it possible to issue a SHUTDOWNCMD command from a slave or only from a
> master?
>
> The scenario is a Dell R740 hypervisor, and I’d like NUT server to run as a
> virtual machine on it, and shut down the host from the virtual machine NUT
> server. So the NUT server VM would issue a shut down command to the slave
> installed on the Dell hypervisor DOM0. I want SHUTDOWNCMD on the Dell slave
> hypervisor to issue the shutdown command to the UPS, as the NUT server master
> will have been shut down by the host slave by then. The other option would
> be to install NUT server in DOM0 on the hypervisor which I’m not excited
> about.
Slight clarification, SHUTDOWNCMD can shut down either a master or a slave
system, but in all of the cases I can think of, only the master system has the
connection to the UPS (that is, the connection to tell the UPS to cut power to
the outlets), and you're right that it is typically triggered by the
SHUTDOWNCMD.
Think of the NUT master system (the one running the driver) as being a
multiplexor for single-user UPS hardware. This makes more sense when the UPS
has a single USB or serial port for control, but it also helps if you have a
network card (running SNMP, for instance). In the case of a network connection,
you might be able to have a non-master system shut down the UPS after the
master system is down, but if you are doing this from NUT, you would need to be
sure that the non-master driver configuration is only used at shutdown time
($DRIVER -k). Most of the NUT example configurations are written assuming the
master stays up longer than the slaves, in order to keep that hardware
connection to the UPS to request the power to be removed.
I have heard of people doing something similar with ESXi, and having NUT
master/slave running on one of the guests. The key is probably having the
master guest VM request a delayed shutdown of both the hypervisor and the UPS.
You will want to make sure your UPS supports the delay values needed to make
this work (for instance, CyberPower UPSes round down to the nearest minute:
https://github.com/networkupstools/nut/issues/432 ), and that the UPS doesn't
have a race condition if the power comes back in the middle of a delayed
shutdown (see NUT FAQ).
> I’m very aware this is a non standard setup. Is there other better ways to
> accomplish this without running something like a Raspberry PI as the NUT
> server master?
>
>
> ___
> Nut-upsuser mailing list
> Nut-upsuser@alioth-lists.debian.net
> https://alioth-lists.debian.net/cgi-bin/mailman/listinfo/nut-upsuser
___
Nut-upsuser mailing list
Nut-upsuser@alioth-lists.debian.net
https://alioth-lists.debian.net/cgi-bin/mailman/listinfo/nut-upsuser
