Nut upsuser - Jan 2011 - Cyber Power USB UPS'.

Greetings:

   I'm evaluating two replacements for UPS upgrades to my dept. Because
the CyberPower OP650 has been retired, we are evaluating the
CyberPower600VA and the CyperPower 1000AVRLCD. We will be evaluating
others.

    Initial configuration and communications are established via the
usbhid-ups driver, the gambit of tests are all passed with the
exception of one. I seem to be failing the power race test. As I
understand it the power race test happens when mains is lost, a system
and ups shutdown are commanded (I simulate this via the upsmon -c fsd
command), but the mains come back online before the UPS cycles power.
The result is the server is left in a hang state waiting for the UPS
to cycle power. What should happen is the UPS cycles power to restart
the server via an open loop timer. This is not happening, the UPS does
not re-cycle the power.

So my setup...

Debian Server (Lenny) running a Linux 2.6.26-2-686 SMP kernel.
NUT version 2.2.2-6.5 (Latest Lenny Stable)

UPS's under test Cyber Power 1000AVRLCD, Cyber Power 600VA.

Our /etc/nut/ups.conf entry looks like this (for the 600VA)

[MyUPS]
       driver = usbhid-ups
       pollfreq = 5
       offdelay=45
       ondelay=70
       #vendorid=0764
       #product=0501
       #desc = "UPS CP600"

The Test:

  First I check to make sure I have communications with the UPS, this is
useless but hey, with a upsc MyUPS at localhost. I get the full telemetry
dump, in this dump I note that my offdelay and ondelay are not being
sent to the UPS. The default values are still there.

  Pull the plug. Within the poll time, I get the system on battery message.

  Execute a usbmon -c fsd, to force the shutdown command.

  After the system has announced it is entering shutdown, I replug in the
mains.

  Now the strangeness begins.

  I get a USB disconnect message for the USB subsystem, immediately
followed by a reconnect message with the same USB port  vendor id you
name it.

   The system seems to rerun a portion of the shutdown script after the
USB restarts the USB connection to the UPS. The final messages.

Shutting down the UPS ...:Network UPS Tools - UPS driver controller 2.2.2
Network UPS Tools: 0.29 USB communication driver - core 0.33 (2.2.2)

No matching HID UPS found
Driver failed to start (exit status=1)
 Shutdown failed. Waiting for UPS batteries to run down.
Will now halt.

True to the system messages, the system does indeed wait for a UPS power
cycle that never comes.

As a sanity check I reconfigure and plug in a Cyber Power OP-1250 and run
the same test with no issues.

Any insights?

Andrew

Citeren tubbiolo op math.arizona.edu:
> Debian Server (Lenny) running a Linux 2.6.26-2-686 SMP kernel.
> NUT version 2.2.2-6.5 (Latest Lenny Stable)
[...]
> Any insights?
Upgrade. We no longer support this version of NUT. Since nut-2.2.2 was  
released, a lot of changes were made to the usbhid-ups driver. Chances  
are that the problems you're seeing now have been solved already (or  
not).

Best regards, Arjen
-- 
Please keep list traffic on the list (off-list replies will be rejected)

Hi All:

   Someone may have already answered this, I get the digest. I found this
posting in the archives, OF COURSE, after I posted my problem.

http://lists.freebsd.org/pipermail/freebsd-questions/2010-August/220187.html

I can't say it explains everything to me. I still am interested in
ya'lls
thoughts. I can't totally say I understand why this is happening.

   If my limited understanding is correct the USB is dropping out, coming
back, and when it comes back init 0 or S90Halt is run. But now the USB
is given a new 'address' and the orig binding via the driver is gone.
So it seems to me the less kludgy fix would be to assign the port
setting in usb.conf to something more akin to reality so this does not
occur.

   I'd rather not muck with the shutdown script as I'll be applying this
to about 50 systems as I upgrade their UPS'.

Andrew