Peter Selinger wrote:
Hi Rob,
perhaps you should start from the beginning. Did you try any of the
stuff that I wrote to you about? These were my instructions:
I admit I haven't done these steps yet as I was hoping to logic my way
through it because doing this test would entail shutting down everything
attached to the UPS and what not. I apologize for that. I'll get to it
this weekend though, and post the results.
No shutdown scripts are mentioned up to this point. We cannot know for
certain what the problem is unless you actually perform these
experiments, and let us know the results.
True, but it doesn't make my statement about the confusion I have
regarding upsmon, upssched, and shutdown scripts any less true. :)
Granted, that's probably a different discussion than the current one.
I'll post the results from the tests this weekend.
Rob
Peter Selinger wrote:
Rob wrote:
What would I be looking for that I didn't see in my testing with my
windows box? Should I be trying to sniff the serial connection to see
if the UPS is for some reason sending a weird signal that NUT is
mis-interpreting? If so, how would you recommend I do that?
I would try the following: plug your computer into a wall outlet, plug
a desk lamp into the UPS. First, start the driver and upsd (but not
upsmon). Monitor the ups variables with upsc. Pull the power, wait 5
minutes or more, see what happens. Are there changes to ups.status?
Does the lamp go off? Normally, the lamp should not go off until the
battery runs out, which should happen some time after ups.status shows
"LB" (low battery). If the lamp goes off before that, there's a
problem.
Then try the same thing again, this time with upsmon running. When the
UPS gets to "LB", then upsmon should start slowly shutting down the
computer. When the computer is in a safe state, the lamp should go off
(the computer will stay on, as it doesn't know it is not plugged into
the UPS).
It could be that something is misconfigured in your shutdown scripts,
for example, perhaps you call "upsdrvctl shutdown" way too early. This
would yank the power from your UPS. In the above setup, you would see
the lamp go off as the computer is preparing to shut down.
-- Peter
Peter Selinger wrote:
That is very odd indeed. Why should the system shut off abruptly? If
the log shows nothing, then it seems that the UPS just turned off the
power.
I recommend testing this while attaching, say, a lamp to the UPS,
instead of a computer. (You can still use your computer to monitor the
UPS, just don't plug it into it!)
-- Peter
Rob wrote:
Upsmon is responsible for monitoring and reacting to power events.
See also the man pages for upsmon(8), upsmon.conf(5), upssched(8), and
upssched.conf(5).
There could be many reasons your Windows machine lasts longer than
your Linux machine on their respective UPSs. Perhaps the battery on
the second one is older, or has lost some of its charge, or is lower
quality... Batteries tend to wear out quite easily, especially if they
are discharged and recharged a few times.
Did you try switching the two machines?
On my system, I only have 1 UPS and I have monitored it in windows and
in Linux. In windows, the same set of systems will properly wait until
the designated time to shutdown (>5 minutes). When I boot into Linux,
my Linux system performs an abrupt poweroff after about 5 minutes. I
notice that UPS has been shutoff, but that the system has not been
shutdown cleanly as the logs do not show a system shutdown and disks
need to be recovered on bootup.
In windows, I monitored the power output and saw that the battery
drained very quickly down to 60%, and then stabilized. I contacted
CyberPower about this and they stated that this is a normal discharge
pattern for SLA batteries. Does NUT possibly detect the power drain
rate and perform an action based upon that? Even if the UPS was sending
a low power indicator, I can't see why NUT wouldn't do a clean shutdown
of the system?
Rob
_______________________________________________
Nut-upsuser mailing list
[email protected]
http://lists.alioth.debian.org/mailman/listinfo/nut-upsuser
