Citeren Thomas Gutzler <[email protected]>:
Thanks for your very tutorial email. It helped a lot. I have moved my master monitor to the nfs machine and setup upsmon on all other machines.
Good. That is the most robust setup, short of having everything powered from one huge UPS.
My master, which is also nfs server now reads: MONITOR u...@localhost 1 monmaster blah master MONITOR u...@localhost 1 monmaster blah master MINSUPPLIES 2 FINALDELAY 5
You'll want to increase the FINALDELAY here to give the NFS clients some time to save their data. Because all systems now have the same FINALDELAY value, they will all start shutting down at the same time (including the NFS server). Something like 60-120 seconds is usually plenty. Since the load on the UPS will rapidly decrease (with systems powering down), usually the UPS will be able to keep the output powered.
All other systems do this: MONITOR u...@monitor 1 monslave blah slave MINSUPPLIES 1 FINALDELAY 5
Since both UPS devices now are equally important to the upsmon master, monitoring ups1 by all slaves will work. But it would be better to make sure they monitor the ups1 or ups2 they receive power from, in order to prevent nasty surprises later on, should your configuration change.
[...]
The halt shutdown script includes code that runs /etc/init.d/ups-monitor poweroff if $INIT_HALT = POWEROFF, which is controlled by /etc/default/halt. There, default setting is HALT=poweroff So it should all work.
*Verify* that it works by running 'upsmon -c fsd' on the NFS server. This will simulate a critical power situation and starts the shutdown sequence for all systems. It's better to try this out at a time that is convenient to you, than to find out it doesn't work when you're not around to fix it.
Best regards, Arjen -- Please keep list traffic on the list _______________________________________________ Nut-upsuser mailing list [email protected] http://lists.alioth.debian.org/mailman/listinfo/nut-upsuser
