[Nut-upsuser] No shutdown on empty battery

[Arnaud Quette]

2012/2/1 Gerhard Strangar <g.s at arcor.de>:
> Arnaud Quette wrote (2012-01-31 12:38):
>
>>> Jan  9 05:37:07 b1 upsmon[67752]: UPS upsb1 at localhost on battery
>>> Jan  9 06:15:19 b1 syslogd: kernel boot file is /boot/kernel/kernel
>> as told above, being on battery is not sufficient to trigger a system
>> shutdown...
>
> The "boot file" line means that the system rebooted.

right, I was not focus enough.

> And after reboot,
> the /etc/kill-power file still existed, so I guess it did not work properly.

/etc/killpower is cleared at upsmon startup, in case it exists.

you have not answered to my question on where "upsdrvctl shutdown" is
called in the process?
as told, on Debian, it's handled by the halt script.

>> a good test to do is:
>> - to shutdown your computer and unplug it from the UPS. You may want
>> to put a light bulb on the UPS instead of the computer,
>> - restart everything, then issue a power failure,
>> - check that NUT sees both OB (on battery) and most of LB (low
>> battery) flags in ups.status.
>> upsmon will also notify you, as you've already seen.
>
> I'll try to do that with a second UPS. But I thought maybe someone else
> is using a Powerware 5 series with nut and he could tell me if it works
> or not.

if you're able to test the development version (aka Subversion trunk),
I've just committed (r3426) support for ups.test.result.
this should help in troubleshooting your issue.

> What I don't understand is, why "low battery" is used for shutdown.
> neither nut nor the UPS know how long it takes to properly shutdown the
> system, so why wait until it might be too late?

see Charles answer.
I also suspect the battery, that do not provide enough runtime WRT the
LB flag publication.

cheers,
Arnaud
-- 
Linux / Unix Expert R&D - Eaton - http://powerquality.eaton.com
Network UPS Tools (NUT) Project Leader - http://www.networkupstools.org/
Debian Developer - http://www.debian.org
Free Software Developer - http://arnaud.quette.free.fr/