[Nut-upsuser] "battery is low" logged

[Charles Lepple]

On Sep 6, 2019, at 6:28 PM, James  wrote:
> 
> I got around to unplugging my UPS and I found it doesn't last nearly as long as estimated.

The estimates are typically less accurate when the load is light (output is showing 5%). Also, as Eyal mentioned, five years is getting old for a lead-acid battery.

> 
> Is the "upsmon[1446]: UPS desktop_ups at localhost battery is low" logged when "battery.runtime:" = "battery.runtime.low: 120"?

To first approximation, yes. Technically, upsmon is looking for the LB flag in ups.status, which is usually sent in a status bit from the UPS. But "battery.runtime.low" is one of the things that triggers LB (the other being "battery.charge" going below "battery.charge.low"). There are also ways for NUT drivers (see "ignorelb") to make that decision in the driver code rather than looking at the UPS status bit, though generally APC seems to get this right.

On Sep 6, 2019, at 7:02 PM, Eyal Lebedinsky wrote:
> No UPS that I owned could tell that the batteries are bad, I find this when the UPS power is
> disconnected and the UPS shuts down soon (often in seconds).

APC, Tripp-Lite, and MGE/Eaton typically do periodic tests that can detect this. Sometimes it isn't on the entry-level models, and sometimes you have to trigger the test manually, but it's definitely out there. (Getting a notification from NUT in time to replace the battery is sometimes another issue entirely, since it isn't easy to simulate battery failures.)

> 
> You can test the condition of the battery by periodically removing the power input and see how long
> it actually runs down to, say, 20%.
> 
> Vendor software usually has a "test" feature, don't know if nut has this.

test.battery* instant commands: https://networkupstools.org/docs/user-manual.chunked/apcs02.html

(I don't know if test.panel.start is exactly the same.)

More detailed status is typically reported in the "ups.test.result" variable: https://networkupstools.org/docs/user-manual.chunked/apcs01.html

(The driver should pick up on this and also include "RB" (replace battery) in "ups.status", but again, this is hard to test.)

James: in your case, this might be due to your APC model not showing the full set of variables over the standard USB HID interface. (Search the web for "APC" and "Modbus" for details.)