[Nut-upsuser] Help needed: LIEBERT PowerSure Personal XT - USB interface

[Alain Williams]

On Fri, Jul 13, 2007 at 10:44:17PM -0400, Charles Lepple wrote:

> Your UPS supports both BelowRemainingCapacityLimit and
> ShutdownImminent, so one of those should trigger the LB condition.

It seems to, it just doesn't give enough time to shut the box down.


> If it's not a sine-wave UPS, then the computer might show up as
> drawing more current when on battery. Load and battery charge are
> usually reported in increments of 5 or 10% anyway, so this shouldn't
> be too much cause for alarm.

Ta

> >Alternatively the resetting of PATH can be done in /etc/sysconfig/ups
> >(as show later).
> 
> Actually, for this issue (and some of the others below), would you
> mind filing a tracker item for this?
> 
> http://alioth.debian.org/tracker/?group_id=30602

Done. ID 304601

> >You need to mention somewhere that the firewall may need to be modified to 
> >allow
> >incoming connections to TCP port 3493.
> 
> The port number is mentioned in the README.

I was following the 'Build and install' in the INSTALL file.
Might be nice to put a one liner about the firewall at about point:
	'10. Start the network server.'.

> >I pulled the plug, got lots of warning messages about the thing being on
> >battery -- after a few minutes it decided to power down but never got there
> >as the UPS powered off before Linux had shutdown properly.
> >Plug the UPS back in and the machine came up without attention.
> >
> >I have had a look at upssched ... looks complicated, why not just add
> >an extra parameters to ups.conf that 'fakes' an imminent power off
> >at the earliest of:
> >* X seconds of the UPS being on battery
> >* the battery.charge being less than Y%
> >* the battery.runtime less than S seconds
> 
> Hmm, that looks complicated too :-)
> 
> I don't think we'd be opposed to including a script that does just
> that, plus some instructions on how to run that from upssched. Bear in
> mind that a lot of UPSes are smart about their LB signal, and will
> send that even if they haven't totally updated the runtime or charge
> level information.

Yes: of course the LB signal would still be used.
The trouble with a time based upssched solution is that once it is
started it can't take account of something like non-usual battery.charge
changes, or the battery not having completely recovered from an earlier outage.


For many users [[read: for me at least :-) ]] it would be simpler to put
a few lines in ups.conf that describe characteristics of the UPS.

[mintups]
    driver = usbhid-ups
    port = auto
    vendorid = 06da
    offat_batterylife = 300
    offat_chargepercent = 50
    offat_batteryruntime = 200

If you have a setup with more than one UPS - it would not work as well;
but I am not sure what would.


> >I attach a new patch against the current RedHat/CentOS 4.5 /etc/init.d/halt
> >This basically checks for /etc/killpower and if it is there executes:
> >        upsdrvctl shutdown
> >
> >I tested it by creating /etc/killpower by hand - it powers off;
> >unfortunately (in my testing) the machine then stayed off, even when
> >I unplugged/plugged in the UPS to the mains.
> >
> >I would like the machine to come back up when power returns, I am
> >not convinced that this setup will do that.
> 
> Please add these to the patch tracker. (I don't have a Fedora or
> RedHat box to test this against, and I would rather give some of the
> Fedora-based developers a chance to look at these patches instead of
> just blindly committing them myself.)

See patch 304610.

I am happy to help maintain this on a RedHat/CentOS system, also test
out any new Liebert stuff.

> Thanks for testing the HID driver!

Thanks for providing it in the first place and putting up with me telling
you how I think you could do it better!

-- 
Alain Williams
Linux Consultant - Mail systems, Web sites, Networking, Programmer, IT Lecturer.
+44 (0) 787 668 0256  http://www.phcomp.co.uk/
Parliament Hill Computers Ltd. Registration Information: http://www.phcomp.co.uk/contact.php
#include <std_disclaimer.h>