[Nut-upsuser] APC Smart-UPS 1500; system shutdown is not initiated.

[Charles Lepple]

On Dec 7, 2013, at 2:38 PM, Christian wrote:

> Dear list,
> 
> I have an APC Smart-UPS 1500 hooked up via the USB port. Recently, when
> we had a power outage, the only entry in the logs was:
> 
> Dec  3 14:43:43 afs1 upsmon[3035]: UPS apc at localhost on battery
> 
> and then the machine seems to have run until the battery was empty. It
> only notices the "battery low" after I turn the system back on:
[...]

> Occasionally, i get these in daemon.log:
> Dec  4 03:11:54 afs1 usbhid-ups[21973]: libusb_get_interrupt: could not
> claim interface 0: Device or resource busy

This isn't great, but I don't think it's related.

It is normal to get a "could not claim interface 0" error once after the USB cable is first plugged into a Linux box. When usbhid-ups starts up, if it gets that error, it has to detach the kernel HID driver. After that point (until the USB cable is unplugged or the box is rebooted), you shouldn't see that error. I don't think you should see it from the libusb_get_interrupt function, either. I'd check the logs here, but the usbhid-ups driver is running on a BSD box at the moment, which doesn't have the same problem as Linux.

How often do you see it? Is there a possibility of another program trying to access the UPS?

> Dec  6 11:50:55 afs1 usbhid-ups[21973]: libusb_get_interrupt: error
> sending control message: Connection timed out

We increased the USB timeout in 2.7.1, so this should go away after upgrading.

These sorts of transient errors are not a problem unless you get several in a row, and the driver should log a different message at that point.

> afs1:/var/log# upsc apc at localhost
> battery.charge: 100
> battery.charge.low: 10
> battery.charge.warning: 50
> battery.mfr.date: 2008/05/22

^ If this date is accurate (and I don't know for sure if our code can reliably update this), then you may want to consider a new battery. Although an UPS battery doesn't see the extreme temperature cycling that a car battery does, it does typically run a few degrees above ambient, and the lead-acid chemistry is only good for 3-5 years of reliable service.

> [...]
> ups.test.result: No test initiated

^ Occasional battery tests are needed to ensure that the calibration values can predict when the battery is about to run out.

Some info is here:

http://forums.apc.com/thread/8669

If you post the output of 'upscmd -l apc', we can try to figure out which calibration command would be the best to try.

-- 
Charles Lepple
clepple at gmail