fix(ci): for old versions of iperf, server process shall be killed manually
When I installed the new powered USB hub in the Faraday cage, I followed the next CI run. I had rebooted both phones and checked their power level.
The next run got stuck again and I saw that on nano there was tens of zombie processes for iperf -s
on nano and the connected phones, the version is 2.0.5 and iperf -s -t 30 does not work (-t is a client-only option).
Dhanuja and I saw that iperf-2.0.10 was installed on the phones as well and I installed 2.0.10 on nano.
But when we tried, we saw that the server file was full of nan% making the IperfServerAnalyze method crash.
So we reverted to 2.0.5 and implemented this fix.
Note that we should discuss what is happening in BIDIR case --> currently there is no -t option for the server.