How to track Network Latency? (Debian)

August 16, 2017 218 views
Networking Debian

Hi,

Question:
How can I track network latency at Debian 8.1 ? I do run a python script 24hours a day, so it can be a program or a python modul solution.
Maybe someone can write a tutorial for that? I only saw a tutorial for "FreeBSD 11" which uses "SmokePing". If SmoePing would solve my problem, can someone write a tutorial how to install and use it on Debian 8.1?
See story for more details:

Story behind it:

I have a Droplet with Debian 8.1.
I do run a python script that does the whole day REST API calls to different services (not more than rate limit). Usually ~30 calls per minute, cause it needs to stay up to date and not all services offer websocket.

Now it happens occasionally that my script raises more timeout errors than usual (==waited 30 seconds for response, but got none). It is not like the connection would be totally gone, it is more like every ~tenth call has a timeout over a timeframe of 5 minutes. After that everything is fine again for a while.

If this happens too often I do ask at the service and at digitalocean support, why the connection is so bad. Usually both are telling me, that the connection is fine. They still ask me to send a ping and mtr test results, which also shows that connection is fine.
But of course I do these tests in a timeperiod where connection is indeed fine. I can't predict if or when the next timeouts will happen. And even if I run mtr the whole day, will that result help if I had 10 timoutes scattered over the whole day? I doubt it.

That's why I'm asking here for help.

1 comment
  • I also heard that "tcpdump" might be a solution (will search for tutorial now...)
    If you have more ideas and at best also tutorial links, it would be very good :)

Be the first one to answer this question.