featureprofiles 100g fr interface config on the dut side

(moving this conversation from #2953 ) In some tests, there's an unusual configuration performed on DUT only for 100G FR ports, e.g.: https://github.com/openconfig/featureprofiles/blob/91796ca845a55876969dec6e9a574c5cc0800738/feature/interface/aggregate/otg_tests/aggregate_all_not_viable_test/aggregate_all_not_forwarding_viable_test.go#L347-L351

It is unclear why is this required.

@dplore suggested that it might be related to the FEC config on the OTG side (https://github.com/search?q=repo%3Aopenconfig%2Ffeatureprofiles+%22Disable+FEC+for+100G-FR+ports+because+Novus+does+not+support+it%22&type=code), but the parameters do not match:

OTG config only disables the fec, while the autonegotiation is still on

https://github.com/openconfig/featureprofiles/blob/22dc0891eabdcf9b9b7bcf11a216c3d734f8f120/feature/gribi/otg_tests/encap_frr/encap_frr_test.go#L898-L903

DUT disables autonegotiation completely

Another question is, if this is needed in all cases to interop with OTG, why do only a few tests use this config?

May 10 '24 00:05 LimeHat

@Ankur19 and @liulk do you have some experience with this issue?

May 10 '24 03:05 dplore

A case in point: looks like it caused an issue for another vendor in #2984

May 13 '24 13:05 LimeHat

@LimeHat are you suggesting the that the regular expectation should be auto-negotiation should work and the deviation should be to disable auto-negotiation and set duplex and port speed?

May 13 '24 17:05 dplore

Yes, I believe so. I'm not aware of any IEEE limitations that would prevent auto-negotiation from working with FR optics.

May 13 '24 17:05 LimeHat

Bumping this up. Any additional feedback? Is there an owner of this code who needs it? If not, I'll open a PR to remove it.

Jul 01 '24 15:07 LimeHat

no additional comments from me. It seems ok to remove. If you send the PR we'll run it to confirm

Jul 01 '24 16:07 dplore