Nut upsuser - Oct 2009 - USB/UPS connectivity issue on Intel DP55KG Motherboard

Hello,

I use the following in my ups.conf:

# USB:
[apc]
   driver = usbhid-ups
   port = auto
   desc = "APC 1500VA"
   pollfreq = 5

On my new motherboard (an Intel DP55KG), every once and awhile I get this:

Broadcast Message from nut at box.com
         (somewhere) at 17:48 ...

Communications with UPS apc at localhost lost


Broadcast Message from nut at box.com
         (somewhere) at 17:48 ...

Communications with UPS apc at localhost established

On my previous motherboard (a DG965WH), the exact same configuration was used
and I never had this issue.

Any hints?

Should I increase the pollfreq to say 30 seconds?

Would this 'workaround' the problem?

Is this a known bug on this chipset?

Justin.

On Tue, 13 Oct 2009, Justin Piszcz wrote:
> Hello,
>
> I use the following in my ups.conf:
>
> # USB:
> [apc]
>  driver = usbhid-ups
>  port = auto
>  desc = "APC 1500VA"
>  pollfreq = 5
>
> On my new motherboard (an Intel DP55KG), every once and awhile I get this:
>
> Broadcast Message from nut at box.com
>        (somewhere) at 17:48 ...
>
> Communications with UPS apc at localhost lost
>
>
> Broadcast Message from nut at box.com
>        (somewhere) at 17:48 ...
>
> Communications with UPS apc at localhost established
>
> On my previous motherboard (a DG965WH), the exact same configuration was
used
> and I never had this issue.
>
> Any hints?
>
> Should I increase the pollfreq to say 30 seconds?
>
> Would this 'workaround' the problem?
>
> Is this a known bug on this chipset?
>
> Justin.
>
>
>
More logs:

Oct 13 18:01:46 p34 upsd[7338]: Data for UPS [apc] is stale - check driver
Oct 13 18:01:46 p34 upsmon[7342]: Poll UPS [apc at localhost] failed - Data
stale
Oct 13 18:01:46 p34 upsmon[7342]: Communications with UPS apc at localhost lost
Oct 13 18:01:50 p34 upsd[7338]: UPS [apc] data is no longer stale
Oct 13 18:01:51 p34 upsmon[7342]: Communications with UPS apc at localhost
established
Oct 13 18:02:48 p34 upsd[7338]: Data for UPS [apc] is stale - check driver
Oct 13 18:02:51 p34 upsmon[7342]: Poll UPS [apc at localhost] failed - Data
stale
Oct 13 18:02:51 p34 upsmon[7342]: Communications with UPS apc at localhost lost
Oct 13 18:02:52 p34 upsd[7338]: UPS [apc] data is no longer stale
Oct 13 18:02:56 p34 upsmon[7342]: Communications with UPS apc at localhost
established

Justin.

On 10/13/2009 03:55 PM, Justin Piszcz wrote:
> Hello,
>
> I use the following in my ups.conf:
>
> # USB:
> [apc]
> driver = usbhid-ups
> port = auto
> desc = "APC 1500VA"
> pollfreq = 5
>
> On my new motherboard (an Intel DP55KG), every once and awhile I get this:
>
> Broadcast Message from nut at box.com
> (somewhere) at 17:48 ...
>
> Communications with UPS apc at localhost lost
>
>
> Broadcast Message from nut at box.com
> (somewhere) at 17:48 ...
>
> Communications with UPS apc at localhost established
>
> On my previous motherboard (a DG965WH), the exact same configuration was
> used
> and I never had this issue.
>
> Any hints?
>
> Should I increase the pollfreq to say 30 seconds?
>
> Would this 'workaround' the problem?
>
> Is this a known bug on this chipset?
Anything unusual showing up in dmesg?

On Wed, 14 Oct 2009, Justin Piszcz wrote:
> Hi,
>
> I have included debugging logs-- see below, is this a nut or kernel/USB 
> subsystem issue?
>
> There are no kernel messages in dmesg/syslog that relate to this event. The
> same exact hardware was used on a different motherboard (DG965WH) without
any
> problems.
>
> Justin.
>
> On Wed, 14 Oct 2009, Justin Piszcz wrote:
>
>> On Tue, 13 Oct 2009, Robert Hancock wrote:
>> 
>>> On 10/13/2009 03:55 PM, Justin Piszcz wrote:
>>>> Hello,
>>>> 
>>>> I use the following in my ups.conf:
>>>> 
>>>> # USB:
>>>> [apc]
>>>> driver = usbhid-ups
>>>> port = auto
>>>> desc = "APC 1500VA"
>>>> pollfreq = 5
>>>> 
>>>> On my new motherboard (an Intel DP55KG), every once and awhile
I get
>>>> this:
>>>> 
>>>> Broadcast Message from nut at box.com
>>>> (somewhere) at 17:48 ...
>>>> 
>>>> Communications with UPS apc at localhost lost
>>>> 
>>>> 
>>>> Broadcast Message from nut at box.com
>>>> (somewhere) at 17:48 ...
>>>> 
>>>> Communications with UPS apc at localhost established
>>>> 
>>>> On my previous motherboard (a DG965WH), the exact same
configuration was
>>>> used
>>>> and I never had this issue.
>>>> 
>>>> Any hints?
>>>> 
>>>> Should I increase the pollfreq to say 30 seconds?
>>>> 
>>>> Would this 'workaround' the problem?
>>>> 
>>>> Is this a known bug on this chipset?
>>> 
>>> Anything unusual showing up in dmesg?
>>> 
>> 
>
> Robert/Nut User List--
>
> Looks like the data goes stale, please let me know if any more
logs/debugging
> is required, is this a kernel/usb problem or a nut issue? I recall a change
> in the USB subsystem a few kernel versions back caused a different issue
(on
> a different UPS, but nevertheless it was kernel related).
The details+logs are here (previous e-mail did not go through, too large for 
the list):

http://home.comcast.net/~jpiszcz/20091014/nut-problem.txt

Justin.

On Oct 13, 2009, at 5:55 PM, Justin Piszcz wrote:
> Hello,
>
> I use the following in my ups.conf:
>
> # USB:
> [apc]
>  driver = usbhid-ups
>  port = auto
>  desc = "APC 1500VA"
>  pollfreq = 5
On NUT 2.4.1 (and some older versions), the pollfreq value determines  
how often usbhid-ups explicitly requests all of the values; however,  
the UPS is still free to send interrupt reports at any time within  
that polling interval, and those reports should be processed more  
frequently.

Typically, interrupt reports are sent by the UPS when status  
conditions change, so the reporting delay for something like a power  
failure should be well below pollfreq.

Was there a particular issue that led you to choose "5" for pollfreq?
If so, what version of NUT were you using at the time?

On Thu, 15 Oct 2009, Charles Lepple wrote:
> On Oct 13, 2009, at 5:55 PM, Justin Piszcz wrote:
>
>> Hello,
>> 
>> I use the following in my ups.conf:
>> 
>> # USB:
>> [apc]
>> driver = usbhid-ups
>> port = auto
>> desc = "APC 1500VA"
>> pollfreq = 5
>
> On NUT 2.4.1 (and some older versions), the pollfreq value determines how 
> often usbhid-ups explicitly requests all of the values; however, the UPS is
> still free to send interrupt reports at any time within that polling 
> interval, and those reports should be processed more frequently.
Ah..
>
> Typically, interrupt reports are sent by the UPS when status conditions 
> change, so the reporting delay for something like a power failure should be
> well below pollfreq.
Ok..
>
> Was there a particular issue that led you to choose "5" for
pollfreq? If so,
> what version of NUT were you using at the time?
Yes, I wanted to be able to see power alerts / events when they were short.
When pollfreq is not set, or if it is set too high (e.g. 30 seconds), I do
not get any e-mail if the power goes out and then comes back on very quickly.

Ok, so is this a bug with the USB chipset used on this DP55KG board or is 
it a problem with nut?  OR did yet another change occur in the kernel usb 
driver, I recall before I had to enable a specific option for other 
problems I used to have on a different UPS 1-2 years ago.

Justin.

On Thu, 15 Oct 2009, Justin Piszcz wrote:
>
>
> On Thu, 15 Oct 2009, Charles Lepple wrote:
>
>> On Oct 13, 2009, at 5:55 PM, Justin Piszcz wrote:
>> 
>>> Hello,
>>> 
>>> I use the following in my ups.conf:
>>> 
>>> # USB:
>>> [apc]
>>> driver = usbhid-ups
>>> port = auto
>>> desc = "APC 1500VA"
>>> pollfreq = 5
>> 
>> On NUT 2.4.1 (and some older versions), the pollfreq value determines
how
>> often usbhid-ups explicitly requests all of the values; however, the
UPS is
>> still free to send interrupt reports at any time within that polling 
>> interval, and those reports should be processed more frequently.
> Ah..
>
>> 
>> Typically, interrupt reports are sent by the UPS when status conditions
>> change, so the reporting delay for something like a power failure
should be
>> well below pollfreq.
> Ok..
>
>> 
>> Was there a particular issue that led you to choose "5" for
pollfreq? If
>> so, what version of NUT were you using at the time?
> Yes, I wanted to be able to see power alerts / events when they were short.
> When pollfreq is not set, or if it is set too high (e.g. 30 seconds), I do
> not get any e-mail if the power goes out and then comes back on very
quickly.
>
> Ok, so is this a bug with the USB chipset used on this DP55KG board or is
it
> a problem with nut?  OR did yet another change occur in the kernel usb 
> driver, I recall before I had to enable a specific option for other
problems
> I used to have on a different UPS 1-2 years ago.
>
> Justin.
>
>
Hi,

I tried removing it, the problem persists, is there anything else I can 
try?  It would be nice to control my UPS in Linux.

Justin.