Open adam-kulagowski opened 4 years ago
@adam-kulagowski i think this is a good idea. Should be optional using a new field. Do you have a PR?
Sorry it is way above my coding skills :/
If You guys have not time for this Feature Request I can ask someone at our team to tale a look but that person is just a C+DPDK programmer with no knowledge of TREX.
Sorry but we wont be able to provide a PR during following months. Can You treat this Issue as a Feature Request?
We are currently testing TREX in environment where:
In result stream stats (which I believe are based on external IPv4.ID) are lost. I've recently discovered that this limitation might be no longer valid for latency stats where we check the Magic at the end of the packet ( https://github.com/cisco-system-traffic-generator/trex-core/pull/489/ )
Since in our case we don't care about the internal IP/UDP payload (SUT is forwarding traffic not terminating it) I was wondering if you could extend this behavior (by adding MAGIC at the end of each packet) for all flow packets (not only the latency).
I'm fully aware that will require to run TREX in software mode and most likely will affect latency but it will give TREX huge flexibility gain in case of tunneling.
If I failed to understand the TREX code in above PR feel free to point that out :)