03-22-2021 01:42 PM
We currently are using FEX (N2K-C2348UPQ-10GE) uplinked to a redundant pair of Nexus 5672UPs.
We are getting around 30% negotiation failure with all of the following scenarios/variables:
We have physically checked/replaced fiber & SFPs. If we move the interface to a different FEX port, same issue. If we move the interface directly onto the Nexus 5K, we do NOT experience this issue (but we lost our redundancy). The vendor has run debug on the negotiation and they wait 40ms for Auto-Neg (which doesnt work), then they wait an additional 40ms for 1000/Full. But that times out. What might the timers be on the FEX ports for Auto-Negotiation/standard negotiation vs the Nexus 5k? Is there a way to speed up the negotiation on the FEX units to fall within these parameters?
03-22-2021 01:48 PM
we need more information is this FEX single home or dual homed? is nexus vPC?
can you share some configuration and FEX outputs
here is some troubleshoot tips :
03-23-2021 07:17 AM - edited 03-23-2021 07:46 AM
This is a dual-homed FEX, using vPCs
Sample::
interface port-channel102
description vpc to Site05-5k-fex102
switchport mode fex-fabric
no lacp suspend-individual
fex associate 102
vpc 102
Would removing the HW queue limit on the FEX interfaces help in this scenario ?
03-23-2021 02:31 PM
i do not see as per the FEX config, we need more information.
we need some information any policies ? on nexus ?
06-23-2021 06:21 AM
what code/image you running? Have you run a bug check?
This should be straightforward based on conf you have on parent 5k. Conf is good.
So, I ain't sure what else it could be, other than maybe you need an image update.
HTH
mohamed
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide