So the plan today was to make a new LXC that could run various collection tools across my network, collect data, and feed it into influx. My first target of course, was the pile of UPS's I own. Right now I have 4 "smart" UPS's, and one dead one, which I just ordered a new battery for.
All my working UPS's are made by TrippLite. Now one of the biggest problem with TrippLite UPS's, is that they have a terrible USB interface.
Bus 001 Device 007: ID 09ae:2009 Tripp Lite
idVendor 0x09ae Tripp Lite
idProduct 0x2009
bcdDevice 0.10
iManufacturer 3 Tripp Lite
iProduct 1 TRIPP LITE UPS
iSerial 2 FW-2473 A
All of my tripplite's have identical product/vendor/serial numbers. This means, you can NEVER plug more than one into the same USB. Worse, they are a little flaky, so even if you think you are smart, and map them by device #, you will lose, because they will disconnect randomly and pick a new dev#. Also, bonus, if you plug them into a powered USB hub instead of the PI directly, they pretty much don't work at all.
OK fine, whatever, can't wire them all to a single PI. I've got them all wired to different hosts now, running ups-nut. I make it so my central VM can connect to all those nut instances, and then find a nifty bit of python that pulls data. After a few hacks of supervisord, and other bits (because this was meant to run in docker, and I hate docker), I fire it up. total fail. Why? because my UPS doesn't present ups.load. What is ups.load? Why that would be the current watt load of the UPS. So basically all I can get off the ups is "charge %" and "not on fire".
Never buy a TrippLite UPS. Just don't. These aren't even the cheap models. They are SMART1500LCD models. Which have the exact same data as the cheap 550W one in the laundry room. Worthless.
Either way, here is my sad pandas graphana dashboard. I'm pretty sure the runtimes are BS.
