Ubiquity stuff has always been flaky with metric accuracy, its commonly mentioned on their forums, and I get the impression they never intended it to be super accurate, just a general overview.

My personal experience with a USG is that under real load it will deprioritize stats, so traffic speeds etc start getting dropped, though I guess thats better than losing network speed just to make a pretty graph

The USG is pretty old at this point, so that's not too surprising. I wonder if their newer hardware suffers from the same issues.

Its true its aging but still works well for a small network. From what I gather in the forums, it might be due to the controller software not so much the specific hardware; the complaints still seem fairly steady.

For example the past year someone reported their printer used less than 30KB but was recorded as 1.6GB; another with a Cloud Gateway (not too old) reporting 17GB of traffic daily for a Kindle consuming less than 1MB

Edit: apparently Ubiquiti quietly EOL'd the USG, so no more security updates. Just bought a UCG Ultra, so far seems a decent replacement, I like that the little screen rotates when mounted upside down

it's hard for me to understand how "pretty old" could excuse something as simple as not incrementing some 64 bit counters correctly.

here's an old post of mine about it. this was a year ago, maybe it's improved but given they were willing to ship this I am not optimistic.

https://nelsonslog.wordpress.com/2023/11/19/ubiquiti-routers...