Rich Brown <notifications@github.com> writes:

> Actually, I want the ability to plot the "best"`ping_host` (that is, lowest latency host), since that's going to display the biggest variance. A few milliseconds of additional bloat added to the 90msec from my house to Fremont is easy to ignore. Tacking a few milliseconds onto 16msec from Cloudflare's DNS shows a significant delta - and worth reporting.
>
> For example, here are ping times for various hosts from my house and (in paren's from netperf.bufferbloat.net in Atlanta):
>
> * flent-fremont.bufferbloat.net ~90msec (~53msec)
> * netperf.bufferbloat.net ~40msec (~30 microseconds - Duh :-)
> * gstatic.com and 8.8.8.8 (Google's all-purpose host and DNS) is ~21-22msec (~15msec)
> * Cloudflare's 1.1.1.1 (anycast address, I suspect) is ~16msec (< 300 microseconds)
>
> So my preference would be to have a way to tell Flent to use ping
> times from another host besides the -H host.

Hmm, that is not easily doable as it is now, unfortunately...

One way to achieve something like it is to add another test parameter
that would override the ping destination of the existing tests. I.e.,
instead of adding a new ping host, you'd just get the ability to
override the one(s) already defined by the test...

Would that work for you?


-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/tohojo/flent/issues/181#issuecomment-530481882