[Nut-upsuser] Will MGE-UPS Ellipse USBS auto shutdown?

Peter Selinger selinger at mathstat.dal.ca
Sun Aug 13 15:47:38 UTC 2006


Daniel O'Connor wrote:
> 
> --nextPart2732627.jUZ38zfOrR
> Content-Type: text/plain;
>   charset="iso-8859-1"
> Content-Transfer-Encoding: quoted-printable
> Content-Disposition: inline
> 
> On Sunday 13 August 2006 16:49, Peter Selinger wrote:
> > Not all UPS hardward and not all drivers support ondelay/offdelay.  In
> > any case, you should let your system shutdown script kill the UPS, at
> > the very end, after all the disks have been mounted read-only, and
> > just before the system is normally halted. This will turn off your
> > computer safely. -- Peter
> 
> Any UPS worth it's salt has ondelay/offdelay (of some form) for 2 reasons, 1)
> without it you can get into a situation where the UPS will not power cycle
> the load if the mains comes back on at the wrong time, and 2) not all
> operating systems re-mount disks read only at shutdown (eg FreeBSD runs
> shutdown procedures then the kernel flushes and un-mounts the disks)

I don't see the first reason. Any UPS, when it receives the "shutdown"
command, should turn off the power for a minimum amount of time before
awaiting AC (just in case the AC power came on before the command was
received). This minimum amount of time (typically a few seconds) is
independent of any settable offdelay/ondelay timer, in fact I don't
see why it needs to be settable.

The second reason justifies a settable offdelay, but not a settable ondelay. 

Several of the newhidups subdrivers don't currently implement
ondelay/offdelay. (It is, however, possible to set shutdown timers
with upscmd). Almost none of the serial drivers support it (at least
"grep offdelay drivers/*c" does not turn up many). 

> As for ondelay/offdelay having the same value - NUT disallows it and makes
> sure that ondelay is longer.
> 
> =-
> Daniel O'Connor software and network engineer
> for Genesis Software - http://www.gsoft.com.au
> "The nice thing about standards is that there
> are so many of them to choose from."
>   -- Andrew Tanenbaum
> GPG Fingerprint - 5596 B766 97C0 0E94 4347 295E E593 DC20 7B3F CE8C



More information about the Nut-upsuser mailing list