[Nut-upsuser] [Bug 915985] Re: usbhid-ups regression (APC BE525-RS)

[Arnaud Quette]

Hi Sergey,

cc'ing NUT users list, since others may be interested in the topic
(for those, follow the bug link below).

2012/1/16 Dmitry Andreychuk <915985 at bugs.launchpad.net>:
> Hi Arnaud,
>
> Thanks for your reply! Attaching output for 2.6.0-1ubuntu3.
>
>
> ** Attachment added: "usbhid-ups_2.6.0-1ubuntu3.log"
>   https://bugs.launchpad.net/ubuntu/+source/nut/+bug/915985/+attachment/2676822/+files/usbhid-ups_2.6.0-1ubuntu3.log
>
> --
> You received this bug notification because you are subscribed to nut in
> Ubuntu.
> https://bugs.launchpad.net/bugs/915985
>
> Title:
>  usbhid-ups regression (APC BE525-RS)

ok, I got it!
first, both of your issues are related.
we reverted in 2.6.1 a change on how we get data from the UPS.
we were previously requesting a very large size, while we now only ask
for the data actual size.

your device is reporting extra garbage data (buggy firmware) when
requesting ("Feature report") battery.charge
(UPS.PowerSummary.RemainingCapacity), while the notified version
("Input report", sent by the device upon changes, like SNMP traps) is
good.

this cause, at libusb level, an EOVERFLOW with 2.6.1+, which is logged
every 30 sec. (polling freq), and cause this data to not be processed.

now, Input report data should take over, and feed battery.charge!
I'd be interested in seeing a debug output of a faulty version
(2.6.1+), that includes a power failure.

the bad news: ATM, I've no solution that can be applied upstream and
distributed.
the good news: well, you already know: keep running 2.6.0 for the time being.

cheers,
Arnaud