James Bensley

I'd like to see VLAN support to as an actual "object" type too, just some general thoughts from our use cases (as a service provider): VLANs are a finite resource...

https://github.com/jwbensley/Etherate/blob/master/speed_tests.c#L435 b_speed_avg is a [long doube](https://github.com/jwbensley/Etherate/blob/master/etherate.h#L236), change to `uint64_t` (N.B. b_speed and b_speed_max as `double`): Then replace: ` eth->speed_test.b_speed_avg += eth->speed_test.b_speed; ` with: ` eth->speed_test.b_speed_avg += (eth->speed_test.b_tx - eth->speed_test.b_tx_prev) `...

Reported by @yskripachov ```C diff --git a/functions.c b/functions.c index a36ece5..faad4f4 100644 --- a/functions.c +++ b/functions.c @@ -2214,7 +2214,7 @@ void sync_settings(struct app_params *app_params, // TX has set ACK mode frame...

This looks like Linux is buffering the frames until `recvfrom()` is called.

Hi @tjhacker Hmm, it is possible but what would the requirement/purpoase be? There is already the well known [`arping`](https://en.wikipedia.org/wiki/Arping) tool which does this. For example; ``` # I can see...

Socket is hitting ENOBUFS, need to ignore this errno rather than dying. stall and carry on.

Add some documentation information about how coalescing may interfere. Ref: https://scholzd.github.io/MoonGen/rate_control.html

Can we use nanosleep instead?

Hi @aapocketz - thanks for reporting this issue and I am very sorry for such a long delay in responding, I have been changing jobs so life is crazy right...

I forgot to mention about the latency... @aapocketz : > I could see how this could maybe measure some sort of packet jitter but I don't see how its possible...