[mfontani]

Indeed, my own raspberry pi was returning weird results with the python-based one; I switched to this one and it is more consistent. I had to add an "--output json" option to it to not have to redo all my tooling around tracking the speedtest results, but forgot to make a pull request to the repo, and I've just done that: same json format as the python-based version.

[milankragujevic]

Thank you! You're awesome!

I used to use speedtest-cli to monitor my Internet speed regularly, in fact every 5 minutes, while also logging DOCSIS signal levels, when debugging connection issues.

It didn't give accurate results, and that caused problems for me.

I wanted to switch to the C++ version, but needed the JSON output to be able to log the data and analyze it.

[walrus01]

a raspberry pi 3/3b+ or whatever (not v4) should by no means be used for any sort of speed test, not only is the wifi weak/slow but the wired ethernet interface is also hanging off a usb2.0 bus.

[Fnoord]

> A raspberry pi 3/3b+ or whatever (not v4) should by no means be used for any sort of speed test

You make a valid point though you could test the speed of the Raspberry Pi in question. As long as you take the caveats you mentioned into account, that's an OK context. Which is the problem with GP's post to begin with.

[mfontani]

My pi is connected directly to the router via ethernet, and the bandwidth available to my home connection isn't enough to saturate it "unfortunately" (~95/20) so this isn't massively a problem for me. YMMV.