Nut upsuser - Feb 2012 - My NUT experience

Hey all, I've had a fairly unpleasant experience getting nut to work.
I like the nut architecture and understand 

1. I check the nut hardware list for a usb model with five star support. Great,
my supplier has an Eaton E Series NV.

2. I run blazer_usb as suggested, tweak nut.conf:
MODE=standalone
and ups.conf respectively:
[eaton]
    port = auto
That's pretty easy so far. It took a little tweaking to get the right magic
in
upsd.users and upsmon.conf. It's at this point I wonder why I should have to
tweak any of this stuff for the simple case of 1 UPS directly connected to 1
machine.

3. Turn off the power to test the configuration. OK, I get an "on
battery"
message. I wait a while for the automatic shutdown to kick in. It never does.
The machine dies uncleanly the the battery eventually runs out. I try again,
this time watching the upsc output. OL changes to OB but never gets to LB.

Some reading of the blazer manpage would lead me to believe that this model
can't/doesn't report the battery.charge and I may need to calibrate
somehow? I
spend several fruitless and frustrating hours trying to tweak these values to
trigger an LB status and a clean shutdown.

4. I never get this to work so start looking at upssched.conf so I can just
trigger a shutdown after say 20s on battery. Finally, rather than have to deal
with all this complexity, I give up and have NOTIFYCMD call upsmon -c fsd
without any timeouts etc.

So finally it works, kind of.
Thanks for listening.
Stephen

TL;DR
* The hardware compatibility table shouldn't list this model (Eaton E NV) as
five stars support or should have a large caveat attached.
* Configuration is still tedious for simple home users. Something is wrong if I
have to a) configure users b) write a script which gets called from upssched and
handle. Please, just a simple SHUTDOWN_AFTER_X_SECONDS_ON_BATTER variable and
handle all the state changes internally.

Hi Stephen,

2012/2/19 Stephen <steve at suasol.com>:
> Hey all, I've had a fairly unpleasant experience getting nut to work.
> I like the nut architecture and understand
>
> 1. I check the nut hardware list for a usb model with five star support.
Great,
> my supplier has an Eaton E Series NV.
>
> 2. I run blazer_usb as suggested, tweak nut.conf:
> MODE=standalone
> and ups.conf respectively:
> [eaton]
> ? ?port = auto
> That's pretty easy so far. It took a little tweaking to get the right
magic in
> upsd.users and upsmon.conf. It's at this point I wonder why I should
have to
> tweak any of this stuff for the simple case of 1 UPS directly connected to
1
> machine.
>
> 3. Turn off the power to test the configuration. OK, I get an "on
battery"
> message. I wait a while for the automatic shutdown to kick in. It never
does.
> The machine dies uncleanly the the battery eventually runs out. I try
again,
> this time watching the upsc output. OL changes to OB but never gets to LB.
>
> Some reading of the blazer manpage would lead me to believe that this model
> can't/doesn't report the battery.charge and I may need to calibrate
somehow? I
> spend several fruitless and frustrating hours trying to tweak these values
to
> trigger an LB status and a clean shutdown.
I'm conscious that E Series NV are not the best nor easiest devices
Eaton is manufacturer.
But it's one of the cheapest, which explains the low level of features.

That said, I can help you setting up the right values, to get the
expected behavior.
> 4. I never get this to work so start looking at upssched.conf so I can just
> trigger a shutdown after say 20s on battery. Finally, rather than have to
deal
> with all this complexity, I give up and have NOTIFYCMD call upsmon -c fsd
> without any timeouts etc.
right, I'm also conscious of this weak point of NUT.
I've, for some years now, planned to complete configuration and to
provide code snippets for upsmon, to allow easy configuration for
specific needs.
Still to come, and depending on the available work force :-/
> So finally it works, kind of.
> Thanks for listening.
> Stephen
thanks for your feedback.
> TL;DR
> * The hardware compatibility table shouldn't list this model (Eaton E
NV) as
> five stars support or should have a large caveat attached.
> * Configuration is still tedious for simple home users. Something is wrong
if I
> have to a) configure users b) write a script which gets called from
upssched and
> handle. Please, just a simple SHUTDOWN_AFTER_X_SECONDS_ON_BATTER variable
and
> handle all the state changes internally.
I hope to finally get rid of these issues... soon.
But having told this for years makes me a bit pessimistic on a
possible availability date.

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/

On Feb 19, 2012, at 9:47 AM, Stephen wrote:
> * Configuration is still tedious for simple home users. Something is wrong
if I
> have to a) configure users b) write a script which gets called from
upssched and
> handle. Please, just a simple SHUTDOWN_AFTER_X_SECONDS_ON_BATTER variable
and
> handle all the state changes internally.
Both of these complaints are valid, but they are at odds with the original
design goals of NUT: providing power control and state information over a
network (where different privilege levels are necessary for local and remote
hosts), and keeping things running as long as possible (which may be shorter or
longer than some arbitrary time-on-battery interval). Personally, I find it hard
to muster up the enthusiasm to streamline the one-node case if it will make it
harder to upgrade one of the 10+ node clusters at work.

We have had several efforts to try and make this easy for the one-node case:
GNOME Power Manager integration, and Augeas configuration file template support.
GNOME Power Manager has been superseded by another project, and not much has
happened with Augeas lately.

It's good to get feedback from users, but it's even better when they can
contribute back to the project. Let us know if you're interested in working
on this.

-- 
Charles Lepple
clepple at gmail