TCP connection latency,jitter,lost measurements?

[orion777]

Please suggest me simple tool to test TCP connection. This is necessary to prove MPTCP operation
- I was trying to use tcping: it works fine in windows, but... what it do in linux?? Seems that it is fully different projects with different functionality, options, etc
- I cant emerge hping on my gentoo pri3 64 bit - some conflicts (I'm not sure)

I need a tool, which, like a typical ping, exports its output in text >>output.txt for future analysis.
THE second PROBLEM IS THAT ACK PACKETS CANNOT BE USED IN MY NETWORK SETUP AS A LATENCY MEASUREMENT!!

[massimo]

Eventually net-analyzer/mtr or net-analyzer/netperf can help you in this case.

[Ant P.]

Flent is good for this stuff.

[szatox]

Maybe iperf?

[mike155]

Don't miss Smokeping.

[orion777]

net-analyzer/mtr is not, because it uses ICMP packets and works similar to dynamic traceroute.
smokeping and flent probabry, are using ACK/NACK packets to estimate latency (The use of ACK/NACK is prohibited in my setup, because it will specify wrong data)

Finally, netperf.. it consists of client and server, so, probabry, it will return real latency as I need! However, I cant understand how to run server If I try to run netperf -L 192.168.1.2 it still returns an error, like "are you sure there is a netserver listening on localhost at port 12865?"
Who knows how to work with this stuff? How to start netserver first on the target machine?
Update: upss.. http://www.golinuxhub.com/2016/09/how-t ... th-in.html

[orion777]

So, netperf was what I was searching for, because it uses "client-server" connection. The only this that is bad is that its output contains only statistical data and you cant see every packet.

Flent is a python-written gui, that uses flent, ping, etc. As to me, it is not good to setup long-term measurement with intermediate data report.

And finaly, sockperf. Seems that it can operate like netperf, but has output per each packet! However, we do not have portage for it.