Nut upsuser - Oct 2007 - MGE Pulsar Ellipse 800 off early or box shutdown late

> If you set lowbatt=70 then the behaviour you are seeing is normal.
Don't
> set lowbatt and you'll get the default manufacturer settings which is
> LOWBATTERY at 30% charge level.
>
> Philippe
Thanks for quick response!

I realized this and forget this setting!
For now I think that my box does not recognize signal about low battery -
I'm only started to read SHUT protocoll of MGE, and do not understand that
is this signal a result of some kind of status request from my box to UPS
or UPS send this signal "unsolicited" - I supose the last idea!
May be I need higher POLLFREQ on "ONBATTRY" situation - about 1..2
sec?!
I'm affraid because packet if I will higher pollfreq, the possibility of
collision is also higher! - what mean if ups sending unsolicited "LOW
BATTERY" packet communication became unsync -ed - nedd to resinchronize
wht take to much time, and ups switch off;(
How to work around? - I'm working on scheduling but it's quite tricky.

tovis wrote:
> For now I think that my box does not recognize signal about low battery -
> I'm only started to read SHUT protocoll of MGE, and do not understand
that
> is this signal a result of some kind of status request from my box to UPS
> or UPS send this signal "unsolicited" - I supose the last idea!
The mge-shut driver will ignore anything sent by the UPS unsolicited.
> May be I need higher POLLFREQ on "ONBATTRY" situation - about
1..2 sec?!
This is not the problem. Your setup seems to work fine when the load is
120W and only fails to do what you expect at a 300W load. This means
that the driver is able to read the status from the UPS and acts
accordingly (by shutting down your system). Unfortunately, at the higher
load, the time to shutdown the system takes too long for the remaining
capacity of the batteries, so the UPS will shutdown itself. You need to
verify a couple of things here:

1) Does the shutdown sequence take too long?
2) How old are the batteries? Could it be they are already worn out (you
can expect this after 3-5 years and even sooner at temperatures above 25
degrees centigrade.
3) Is the load really 300W?

A MGE Pulsar Evolution 800 should have a runtime of about 10 minutes
with a load of around 300W, so if it is significantly less you should
really checkout item 2 (and maybe 3). Forcing a battery check may be a
good idea.

You didn't mention which NUT version you're using. If it is anything
recent (nut-2.0.5 or higher) you might want to try the 'newmge-shut'
driver, which is an improved version.

Best regards, Arjen

> 1) Does the shutdown sequence take too long?
Is it about 15 - 20 seconds (from usual system message about shutdown).
> 2) How old are the batteries? Could it be they are already worn out (you
> can expect this after 3-5 years and even sooner at temperatures above 25
> degrees centigrade.
It was renewed month later - I use it about 3 years, but I do not use any
software for shutdown - it is a quite old restance:)
> 3) Is the load really 300W?
Good question! - on state "ONLINE" I have status report from cgi-nut /
apache2 that load is about 70%, on state"ONBATTERY" reports that is it
about 115 .. 120% (what I'm do not understand what is the real).

Debian 4.0 "Etch" including package nut 2.04-4 - it is includes
mge-shut
and mge-utalk - "utalk" does not work for me - too many
"stale" messages.
Of course I can try to "backport" "newmge-shut" - some how:)

I think the best solution for me, is to confugure the upssched, write some
scripts and start shutdown procedure at point when ups "ONBATTERY"
state
is recognized. I have been checked that setting of the parameter
"ups.delay.start" is working, but some what strange what measure it
(sec,
10 sec or else). Could some one give me a good, step-by-step example (the
best what I found is for RedHat and nut version 1.xx?

Thanks to all
  tovis

Overload? OK to be save I will measure it exactly.
Unfortunatelly, I have experiments so many configuration options and
environment (60W lamp, 3*40W lamps and 300W heater) that I can mess some
of them. Also I was experienced two times that everything is goes well, is
it always with starting charge of batteries more than 95%, and ups itself
was restarted (unplug line and batteries together). I think that my
problems also coming from somewhat buggy firmware;( Some times I have run
to problem because of I do not know waht parameters are mean exactly - I
have set "ups.delay.shutdown" (it is always coming up as -1 using upsc
command) and my usp switched of on that time - for me it was a surprize.
More clues, on 300W load, "battery.charge: 86", "ups.load:
68" I have
"battery.runtime: 147" - I think it means about 2 minutes - should be
enough to wheel the shutdown procedure on two boxes.
I repeat I do not need high availability - I need a safe shutdown and wait
to charge up batteries, to prepare for next time - if it takes an hour
than it could have it.
My approach is shutdown as quickly as possible on power failure and come
up as late as possible after line is coming back - usually it is unsafe to
use it after failure at least 10..15 miutes;(
Thanks
  tovis