[Nut-upsuser] MGE Evolution and programmable outlets configuration

[Arnaud Quette]

Hi Marco,

just a (very) quick followup...
the support being overloaded (so do I but), I'll answer to your
questions to speed up the process ;-)

Le mercredi 10 décembre 2008 à 21:52 +0100, Marco Chiappero a écrit : 
> Arnaud Quette ha scritto:
> > [a quick walk through the ml, still from the UDS/Googleplex...]
> > 
> > Hi Marco,
> 
> Hi Arnaud, good to read you!

thanks ^_^

> >> A digression about the fan: it's really annoying not just because it's
> >> louder than my 100W computer with 4 fans running (yes, it is), but also
> >> because there's no need to keep it spinning all the time in a *small*,
> >> *650* UPS generating just a *few* watts during normal use (measured with
> >> a watt-meter and both the metal cage and the exhaust air are indeed
> >> cold). Replacing it with a quieter fan, although fine for the UPS, means
> >> void warranty. I'd like to have a little more attention on this aspect,
> >> maybe using a thermoregulated fan, when paying 300 euro.
> >> But I should address these critics to MGE rather than the NUT mailing
> >> list, I know. However, beyond this issues, the UPS has proven to be a
> >> good product, good-looking too :)
> > 
> > you've been heard ;-)
> > the point you might be missing is that the evolution range is a line
> > interactive one, but more for small servers.
> > and these are generally in some server rooms.
> > for something more SOHO focused (and so less noisy), you should have
> > opted out for Ellipse MAX (still line interactive, with 2 outlets
> > groups too)
> 
> Well, I knew about the Ellipse MAX but I preferred the Evolution, which
> is pure sine wave. Beyond this you are right, but the basic idea is not
> to cool something that don't need to be cooled. For example the APC
> Smart-UPSs (really similar to the Evolution line, line-interactive, sine
> wave) 750VA and 1000VA are totally fanless. I'm sure that the Evolution
> 850 and 650 models can run fanless too, at least during normal use
> (9-12W don't require an always spinning fan or a fan at all). Moreover I
> suppose that nowadays low power UPSs are widespread in offices as much
> as in server rooms, expecially the tower versions.
> I can now ask you why the Evolution is using two different fan speed if
> it supposed to stay in a server room? And, about the heat, where is the
> difference, when not running on batteries, beetwen the Ellipse MAX and
> the Evolution, fan excluded? Why is the first one fanless while doing
> the same thing, with the same tecnology? ;)

ok, I talked with the right people, just to be sure that I won't make
nonsense.

There is a small difference between Ellipse MAX and Evolution that
justify the need of the fan. Even more if you consider that the 2nd are
(were) more targeted at server rooms...

now, you're right that things are evolving, and that we more and more
see that kind of units under desks...

about the 2 fan speeds, and the fact that these are always running:
- the 2 speeds are there to address online/trim/boost for the first, and
onbattery cases,
- the fact that it's never stopped is partly due to the lack of
temperature sensor to act smartly, and to be sure that the fan will
*always* be able to be restarted.

> Now, let's stop thinking of the fan, I love my Evolution and I like your
> presence here. One more reason to prefer MGE. :)

thanks ;-)

> > can I possibly contact you next week, when I'm back to MGE, to get
> > some more feedback?
> 
> Sure!

sorry for the lag.

> > now about the Powershare, these are things that I should have done for years...
> > but since I've taken over the project leadership, there are far too
> > many things I've not been able to give time to.
> > now, the situation has evolved since my Eaton'isation. I'll have more
> > and more time, and possibly somebody to work with me (but the crisis
> > might void this point in the short run though).
> 
> (I know, in these weeks many projects are being delayed or stopped)
> I have a couple of questions for you.
> 1) It's not clear to me whether the UPS is going to reboot or not when
> mains is back again, if it has not been shutted down. Let's consider the
> VoIP phone example, if I drain as much as I can my batteries (with the
> deep discharge protection on) and the UPS is not shutted down by any
> computer before it stops, is it going to reboot when power is restored?
> I suppose it's not going to reboot in case the power is restored before
> the UPS is dead, right?

first note: your load will be shutdown cleanly so (which sometimes is
not a problem at all (at least not a SW issue), so I guess this is your
case...)! 

To answer your question, yes the UPS will restart when the power is
restored.
This is linked to the below HID data (the raw output you'll see when
starting usbhid-ups in debug mode, level 3 min. iirc):
UPS.PowerConverter.Input.[1].AutomaticRestart

This one default to 1, which means "turn on when power is restored".
It's available since some customers might want to turn it off for
various reasons...

> 2) The NUT documentation states that, usually, the event triggering the
> computer(s) shutdown is the "low battery" state (or a timer when using
> upssched). Now I see in psp (for windows, since there is no psp amd64
> for linux :P) that it is possible to call the system shutdown on the
> basis of (3) differen criteria. Moreover the "battery charge" criteria
> seems not to be related with the low battery precentage (the one
> activating the low battery alarm). Is this right? If so, why under NUT
> is the "low battery" state the only one setting/criteria triggering the
> system shutdown? It would be comfortable to start the power down
> procedure when a certain battery charge is reached (not necessarily the
> low battery percentage) or when there are still X seconds of runtime
> available. The good thing is that the same method can be applied to the
> switchable outles. For example I can set the UPS to cut the power on the
> outlet.1 when the batteries are at 55% and have NUT starting the
> shutdown procedure when the charge is lower than 70%. Or I can start the
> system shutdown at 70% and before halting set outlet.1.delay.shutdown.
> Unfortunately this require the system(s) to be aware of the outlet 
> it(they) are using, and is going to complicate the NUT architecture, but 
> such behaviour would make it *really* powerful! The "low battery" state 
> can be still useful, to make systems shutdown as fast as they can,
> eventually calling the UPS shutdown too. Maybe it can be used as event
> for the last computer running, the one controlling the UPS, for example.
> However I don't know if this can be achived easly, but maybe it sound
> interesting and something can be done... just let me know your opinion!
> Many thanks for your attention and thanks for this chat.

these 2 points (more default shutdown options as in PSP, and more /
smarter outlets options for shutdown handling) are part of the TODO
list, or at least part of the discussion/thoughts underway.
I guess / hope that I will soon have even more time to work on these NUT
subjects in the coming time (check next week's Christmas mail for more
info...)

I hope to have answered your questions.

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