Nut upsuser - Sep 2014 - Nut 2.7.2 on OpenBSD 5.6 with APC USB UPS

On 09/11/14 21:43, Charles Lepple wrote:
> On Sep 11, 2014, at 9:48 PM, Stan Gammons <sg063015 at gmail.com>
wrote:
>
>> I have an APC USB UPS that was working on OpenBSD 5.5 with NUT 2.7.1  
When I installed the latest 5.6 snapshot with NUT 2.7.2 the UPS no longer sends
status messages when it loses line power.  I asked over on the OpenBSD ports
list and so far haven't had any luck getting this to work again.  I disabled
upd at boot as was suggested and that didn't change anything except stopping
the random upd detached messages.
> Looking at the diffs between NUT 2.7.1 and 2.7.2 (including the OpenBSD
patches), I don't see any changes in NUT that might cause this.
>
> Your ProductID of 0002 is old enough not to have the broken interrupt pipe
firmware, but that expects the OS to handle USB Interrupt In transfers properly.
Otherwise, it will take up to 30 seconds (pollfreq) to re-fetch all of the
variables that the UPS should provide through the Interrupt In pipe.
I see a libusb_get_interrupt message in the log. Could that be the problem?
>
> I would try starting the driver with "-DDD", and capture that to
a log:
>
> /usr/local/bin/usbhid-ups -DDD -a apcusb 2>&1 | tee /tmp/apcusb.log
>
> Let it run for about two minutes, then ^C, compress the log and mail it
back to the list.
>
> You can also try reducing "pollfreq", although the reason for
this was to prevent the UPS from tripping over itself by sending too much data.
The "ups.status" bits should be polled in the fast polling cycle (2
seconds versus 30). Hopefully the log will have some error messages to explain
why that is apparently not happening.
>
Here's the log file.  Hopefully it will attach Ok.


Stan
-------------- next part --------------
A non-text attachment was scrubbed...
Name: apcusb.tar.gz
Type: application/gzip
Size: 16641 bytes
Desc: not available
URL:
<http://lists.alioth.debian.org/pipermail/nut-upsuser/attachments/20140911/1a100adc/attachment-0001.bin>

On Sep 11, 2014, at 11:04 PM, Stan Gammons <sg063015 at gmail.com> wrote:
> I see a libusb_get_interrupt message in the log. Could that be the problem?
Yes, it seems related.

   1.126896	upsdrv_initinfo...
   1.126906	upsdrv_updateinfo...
   1.127023	libusb_get_interrupt: Function not implemented
   1.127030	Got 0 HID objects...
   1.127036	Quick update...

Come to think of it, this might also depend on how OpenBSD provides libusb-0.1.
What version of libusb-compat do you have?

However, checking the rest of the log, it seems like there are no errors during
the "Quick update" portion, which should provide AC status updates
(among others) every two seconds.

Can you can do a test where you power off the AC to the UPS? (I'd advise
using a power strip, PDU, or circuit breaker to keep the ground connected,
rather than pulling the plug completely.)

If so, same test, but "-DD" should suffice, and record the time (first
column of the log is seconds since the start) when you toggle the power. Let it
go for at least 30 seconds, so  it will eventually show up in the "Full
update".

The OB/OL status is controlled by this line:

 [time]	Path: UPS.PowerSummary.PresentStatus.ACPresent, Type: Feature, ReportID:
0x07, Offset: 2, Size: 1, Value: 1

-- 
Charles Lepple
clepple at gmail

On 09/11/14 22:34, Charles Lepple wrote:
> On Sep 11, 2014, at 11:04 PM, Stan Gammons <sg063015 at gmail.com>
wrote:
>
>> I see a libusb_get_interrupt message in the log. Could that be the
problem?
> Yes, it seems related.
>
>     1.126896	upsdrv_initinfo...
>     1.126906	upsdrv_updateinfo...
>     1.127023	libusb_get_interrupt: Function not implemented
>     1.127030	Got 0 HID objects...
>     1.127036	Quick update...
>
> Come to think of it, this might also depend on how OpenBSD provides
libusb-0.1. What version of libusb-compat do you have?
# pkg_info
libltdl-2.4.2p0     GNU libtool system independent dlopen wrapper
libusb-compat-0.1.5 libusb-0.1 compatibility layer for libusb1
libusb1-1.0.9p8     library for USB device access from userspace
nut-2.7.2p2         UPS monitoring program supporting many brands
quirks-2.11         exceptions to pkg_add rules
radeondrm-firmware-20131002p0 firmware binary images for radeondrm(4) driver

>
> However, checking the rest of the log, it seems like there are no errors
during the "Quick update" portion, which should provide AC status
updates (among others) every two seconds.
>
> Can you can do a test where you power off the AC to the UPS? (I'd
advise using a power strip, PDU, or circuit breaker to keep the ground
connected, rather than pulling the plug completely.)
>
> If so, same test, but "-DD" should suffice, and record the time
(first column of the log is seconds since the start) when you toggle the power.
Let it go for at least 30 seconds, so  it will eventually show up in the
"Full update".
>
> The OB/OL status is controlled by this line:
>
>   [time]	Path: UPS.PowerSummary.PresentStatus.ACPresent, Type: Feature,
ReportID: 0x07, Offset: 2, Size: 1, Value
UPS.PowerSummary.PresentStatus.ACPresent remains at 1 when I turn the 
power off on the power strip the UPS is plugged into. Here's a 73 
seconds long log with the power turned off.


Stan

-------------- next part --------------
A non-text attachment was scrubbed...
Name: apcusb2.tar.gz
Type: application/gzip
Size: 6872 bytes
Desc: not available
URL:
<http://lists.alioth.debian.org/pipermail/nut-upsuser/attachments/20140912/41726348/attachment.bin>