[Nut-upsuser] Nut-upsuser Digest, Vol 45, Issue 13
Arjen de Korte
nut+users at de-korte.org
Sun Mar 15 20:38:41 UTC 2009
Citeren Bob Blackwell <rc.blackwell op yahoo.ca>:
> From what I can tell three's only one active driver. PS shows;
>
> PID USER COMMAND
> 2045 root /ffp/bin/usbhid-ups -a APC_UPS
> 2047 root /ffp/sbin/upsd
> 2049 root /ffp/sbin/upsmon -u monuser
> 2050 nutmon /ffp/sbin/upsmon -u monuser
Could it be that you're running virtualization software and that
you're running NUT on one of the slaves?
> Running usbhid-ups -u root -DDDDD -a APC_UPS>debug.log 2>&1 never
> stops without a ^C. The log shows many disconnect and re-connect
> attempts.
Usually this either means that your USB system isn't setup properly
(or you're in a really 'noisy' environment) or that some other process
is also attempting to claim the same USB device. This will lead to an
endless mess of reconnection attempts for both drivers. USB doesn't
really handle this well if you don't use the kernel (HAL) to startup
drivers, but instead talk to them directly like we do. We do have HAL
addons that would not suffer from this, but these usually are only
worth using if you have a graphical desktop.
> I'm rather green at Linux thus it'll likely take me a couple
> of days to determine if there's a hotplug/udev issue. Until then I'll
> refrain from posting debug's output.
It looks like this might be caused by an (unexpected) use of signed
chars on your system. The HID parser really doesn't cope with that
properly in nut-2.4.1 and before. I have attempted to work around that
in the latest version in the SVN trunk. I could be that the problem
with the strange readings, is that the logic used to calculate the
exponent is off by a factor of 10^256 due to problems with the
interpretation of (signed char) values as (unsigned char) on your
system.
If you post debug output, make sure to include the report descriptor,
which would be the most valueable part, together with the initial dump
of the variables.
Best regards, Arjen
--
Please keep list traffic on the list
More information about the Nut-upsuser
mailing list