Drop Rate in both stateless and stateful

[Dan N]

I am testing an probe with Trex > Network Data Broker > probe

On Trex i have two 40Gbps NICs and two 100Gbps on NDB.

I am testing in stateless mode with: start -f stl/bench.py -m 100% --port 0 1 -t size=590,vm=var1

[Besart Dollma]

Both ports are transmitting but none is receiving. That explains the drop rate.

Is this connected in a loopback?

[Dan N]

Hi Besart, source mac addresses are from 40Gb interfaces(Trex) and destination from NDB 100Gbps interfaces.  

[Dan N]

NDB is getting the traffic tho.

On Friday, January 29, 2021 at 4:16:52 PM UTC+1 besi7...@gmail.com wrote:

[Besart Dollma]

Hi, 

I am not sure you understood me. The NDB gets the traffic and that is fine but is the traffic being redirected back at TRex? 

Think about it this way. If port 0 of TRex sends packet x and port 1 doesn't receive that packet, TRex considers it dropped.

So let's try something at a lower scale and see that the packet that is sent from one port is received at the other one.

Thanks, 

Bes

[Dan N]

Hi,

the probe we are testing is actually getting the traffic from TRex, meaning that NDB doesn't send traffic back to TRex.

[Besart Dollma]

Hi, 

Take a look at the doc, both endpoints of the traffic should be TRex.

https://trex-tgn.cisco.com/trex/doc/trex_astf.html

Thanks

[Dan N]

I have tried a loopback connection and no luck...

R,D

[Besart Dollma]

Hi,

Please consider the steps like in the doc.

https://trex-tgn.cisco.com/trex/doc/trex_manual.html#_configuring_for_loopback