[Nut-upsuser] Nut fsd command ignored: bcmxcp_usb driver, Powerware 5105 UPS

[Stuart Rackham]

Many thanks to all who helped out on this one, the problem turned out to 
be a misunderstanding on my part:

The Ubuntu 7.04 /etc/init.d/halt script only initiates the 
/etc/init.d/ups-monitor poweroff command if the shutdown -P (poweroff) 
option is used (or -h with /etc/halt/default set to poweroff) -- I was 
using the shutdown -H (halt) option so /etc/init.d/ups-monitor poweroff 
  never got executed.

A couple of observations regarding the bcmxcp_usb driver and 3105 UPS:

The bcmxcp_usb driver turns the UPS off immediately (at the same time 
the PC is powered off by shutdown) irrespective of the setting of the 
driver 'shutdown_delay' parameter. I'm just guessing that the driver 
forces the UPS off when it's terminated by the dying kernel. The only 
way I could get the driver to observe the 'shutdown_delay' parameter and 
not to turn the UPS off immediately was to execute the 'shutdown.return' 
instant command manually (the 'shutdown.stayoff' command turned off the 
UPS immediately):

   # upscmd -u admin -p mypass pw3105 at localhost shutdown.return

This left the following message in `/var/log/daemon.log`:

   Aug  2 13:33:30 kea bcmxcp_usb[4880]: Going down in 120 sec


Cheers, Stuart


Arnaud Quette wrote:
> 2007/7/26, Stuart Rackham <srackham at methods.co.nz>:
>> Hi
>>
>> I'm having a  problem with the bcmxcp_usb driver FSD command: everything
>> seems to work apart from:
>>
>>    upsmon -c fsd
>>
>> Environment: Ubuntu 7.04, nut 2.0.5-1, nut-usb 2.0.5-1, Powerware 3105
>> UPS.
>>
>> upsd appears to send the FSD command to the UPS, upsmon then runs my
>> NOTIFYCMD (which halts my PC) but the UPS just doesn't turn off.
>> Here are the /var/log/daemon.log entries:
>>
>> Jul 24 23:13:16 kea upsmon[5226]: Signal 10: User requested FSD
>> Jul 24 23:13:16 kea upsd[5223]: Client monuser at 127.0.0.1 set FSD on UPS
>> [pw3105]
>> Jul 24 23:13:16 kea upsmon[5226]: Executing automatic power-fail shutdown
>> Jul 24 23:13:16 kea upsmon[5226]: Auto logout and shutdown proceeding
>> Jul 24 23:13:21 kea upsd[5223]: Host 127.0.0.1 disconnected (read 
>> failure)
>>    :
>> Jul 24 23:13:28 kea upsd[5223]: Signal 15: exiting
>>
>> I might have suspected the UPS or USB interface/cable if not for the
>> fact that the shutdown.stayoff instant command works -- turns ups off
>> immediately and that notifications are returned from the UPS.
>>
>> The bcmxcp_usb driver reports the following status:
>>
>> # upsc pw3105 at localhost
>> driver.name: bcmxcp_usb
>> driver.parameter.port: auto
>> driver.parameter.shutdown_delay: 120
>> driver.version: 2.0.5
>> driver.version.internal: 0.11
>> input.phases: 1
>> ups.firmware: 00.80
>> ups.model: POWERWARE UPS    500VA
>> ups.power.nominal: 500
>> ups.serial:
>> ups.status: OL
>> ups.voltage.nominal: 240
>>
>> One peculiarity -- even though the test.battery.start instant command
>> is listed it puts this in the log:
>>
>> Jul 24 23:08:11 kea bcmxcp_usb[5221]: [test.battery.start] not supported
>>
>> Any help would be appreciated.
> 
> Debian, and derivatives such as Ubuntu, declares a generic handler in
> the halt script:
>        # See if we need to cut the power.
>        if [ "$INIT_HALT" = "POWEROFF" ] && [ -x /etc/init.d/ups-monitor ]
>        then
>                /etc/init.d/ups-monitor poweroff
>        fi
> 
> This works for nut and apcupsd...
> now, I've asked a long time ago to have libusb in /lib (instead of
> /usr/lib) just to be sure that the lib is still reachable at shutdown
> time (otherwise, the driver wouldn't ever fire up!)
> 
> the strange point is that we have validated it again this week (a new
> MGE PSP release is underway) on Ubuntu (dapper and feisty), using nut
> 2.2.0 and it (the UPS shutoff / poweroff) worked! But I admit there
> might be an upstart (the init replacement) specific out there...
> 
> Is the shutoff working otherwise (ie calling upsdrvctl -k (CAUTION: no
> sensitive load on the UPS; only light bulbs or alike))?
> 
> A (temporary) workaround, if your unit provide a sufficient delay, is
> to have the SHUTDOWNCMD in upsmon.conf pointing to a script calling
> upsrw / upscmd to launch the delayed shutdown with enough time to
> shutdown the system, and then call the shutdown / halt / whatever
> command...
> 
> Arnaud