On Wed, 26 Oct 2005, Arnaud Quette wrote: > > The poweroff action for /etc/init.d/ups-monitor (symlinked to nut) needs > > to run at S90halt, to power off the load. > > > > At that time, everything is unmounted or mounted readonly, etc. So there > > is no way a serial-line ups can be commanded by nut to poweroff the load > > (there is no /var with a socket, no running UPS driver...). > > > I've already made (upstream) the needed changes in NUT 2.0.3-pre1: > >From the changelog: > > 2.0.3-pre1 > > > >Thu Aug 16 10:51:17 UTC 2005 / Arnaud Quette <[EMAIL PROTECTED]> > >... > > - drivers now don't chroot to statepath if called for power off > > (using -k). This avoid unneeded failure to poweroff the UPS if > > /var is umounted (thanks to Gaspar Bakos).
/etc/init.d/nut stop will have run then. Do the drivers run without the need for any other nut component running in memory? I fixed this on my system by modifying /etc/init.d/nut to initiate the UPS shutdown when needed. The UPS gives the load 5 minutes to finish shutting down before it powers it off. Are there immediate-poweroff UPSes that don't support this delay? Those are not useable with Linux anyway, as the kernel will not have marked RAID arrays clean and read-only at S99, nor will it have flushed drive caches. That happens at final system powerdown. -- "One disk to rule them all, One disk to find them. One disk to bring them all and in the darkness grind them. In the Land of Redmond where the shadows lie." -- The Silicon Valley Tarot Henrique Holschuh -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
