Hello
On 06/14/2017 03:32 PM, Arnaud Quette wrote:
On Jun 7, 2017, at 5:47 AM, Manuel Wolfshant <[email protected]
<mailto:[email protected]>> wrote:
>
> If that matters, the OS is a fully updated CentOS 6.9
and this (latest stable ) version of nut was packaged by me.
The problem appears on any of the USB ports ( well, I tried
the 2 in front and one in the back of the server ).
>
> lsusb -v reports:
...
> wDescriptorLength 549
> Warning: incomplete report descriptor
> Report Descriptor: (length is 9)
> Item(Main ): (null), data=none
Because both NUT and lsusb are having trouble retrieving the
HID Report Descriptor, I think the problem is at a lower
level: probably between the UPS, the kernel, and the USB HCI.
The archives have a number of unresolved emails about the 5E
and "broken pipe" errors.
Probably worth checking with Eaton, too.
Charles is right in both the fact that the issue is (or at least
seems to be) upstream to NUT, and also that it's worth checking
with "Eaton"
I've approached "them" ( apparently my request for supported
landed at after-sales support/Romania ) as soon as Charles
replied. Unfortunately the "dialogue" rather stalls, they seem to
have a policy to not send more than one email every 3 days.
Leaving aside that after telling them that I am using an *USB
*unit (and providing USB-related logs ) they wanted to know if I
was using *SNMP*.
no comments ;)
.... still no reply from them ...
Could you please tell me your kernel
[root@belgrade ~]# uname -r
2.6.32-696.3.1.el6.x86_64
that may be part of the issue... long time I've not tried the 2.x
series, not sure how it goes nowadays.
btw, any interesting "usb" messages from your syslog?
also, would you be able to test with a 3.x or 4.x, at least to see if
that improves / solves the issue?
It's a bit tricky. I tried to use the kernel-lt package from the elrepo
repository ( incidentally I am also part of the elrepo team but I am in
charge with other packages, not the kernels ) and failed miserably. For
reasons unknown to me ( and not logged ) the system failed to boot and a
colleague from that office had to manually make it use the stock CentOS
kernel ( typing what I was dictating to him over the phone .. )
Due to a complete lack of logs, I have absolutely no idea what happened
and since the machine is critical for the activity there, I am a bit
reluctant to try again
Would you also be able to test some github code?
We have the libusb-1.0 branch that provides both libusb 1.0
support (interesting to test to see if the problem still happens)
along with few other improvements (though these should not help
for your issue):
https://github.com/networkupstools/nut/tree/libusb-1.0
<https://github.com/networkupstools/nut/tree/libusb-1.0>
I can and I will test ( tomorrow, probably ).
I just packaged and tested it ( see below some comments, not important
for my issue here )
Unfortunately there is no change in the output:
[root@belgrade ~]#u root -x explore -x vendorid="0463" -a eaton
Network UPS Tools - Generic HID driver 0.42 (2.7.4.1)
USB communication driver (libusb 0.1) 0.33
0.000000 [D1] debug level is '4'
0.012088 [D1] upsdrv_initups...
8.248213 [D2] Checking device (0463/FFFF) (002/004)
9.249069 [D2] - VendorID: 0463
9.249094 [D2] - ProductID: ffff
9.249129 [D2] - Manufacturer: unknown
9.249135 [D2] - Product: unknown
9.249140 [D2] - Serial Number: unknown
9.249145 [D2] - Bus: 002
9.249151 [D2] - Device release number: 0001
9.249156 [D2] Trying to match device
9.249199 [D2] Device matches
9.249209 [D2] failed to claim USB device: could not claim
interface 0: Device or resource busy
9.249373 [D2] detached kernel driver from USB device...
9.249394 [D3] nut_usb_set_altinterface: skipped
usb_set_altinterface(udev, 0)
9.249991 [D2] Unable to get HID descriptor (error sending
control message: Broken pipe)
9.250002 [D3] HID descriptor length (method 1) -1
9.250009 [D4] i=0, extra[i]=09, extra[i+1]=21
9.250017 [D3] HID descriptor, method 2: (9 bytes) => 09 21 10 01
21 01 22 25 02
9.250023 [D3] HID descriptor length (method 2) 549
9.250028 [D2] HID descriptor length 549
9.250515 [D2] Unable to get Report descriptor: Broken pipe
Comments on the new tree:
a) there seem to be some missing files in the tree:
- some bits for augeas, devs and udev:
configure.ac:1526: required file `scripts/augeas/nutupsconf.aug.in' not
found
configure.ac:1526: required file `scripts/devd/nut-usb.conf.in' not found
configure.ac:1526: required file `scripts/udev/nut-usbups.rules.in' not
found
- all man pages
b) on top of that, I did not manage to include the manpages in the final
rpms because the new configure script insists on using tools not
available on CentOS 6 ( a newer asciidoc, for a start ) and I was too
lazy to just include the pages as they are. Why did you change that
towards 2.7.4, oooh why ? It was working sooooooooooooo fine....
FWIW, I modified the spec to create a separate subpackage for the
augeas lenses/modules -- those did not seem to exists in 2.7.4. If
anyone wishes to toy with the spec or rpm packages I will be happy to
share them.
Regards,
manuel
_______________________________________________
Nut-upsuser mailing list
[email protected]
http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/nut-upsuser
