Was wondering what other experiences have been in the field with dual-homed FEX 2248s against two vPC peer-enabled 5010s.
We're using this configuration on the latest 4.2(1)N2(1) code and seeing 5-10 second failover times when "pulling the uplink" on our FEX physical uplink to 5010-A. Surprisingly, the same test when re-connecting and then pulling the uplink to 5010-B is less than 1 second.
Each 2248 has a single 10Gb connection back to each 5010 (total of two per 2248) - and at the N5K level each 5010 is configured to have a single-10Gb port-channel (vpc-enabled) back to the FEX.
Is there a STP/BA issue going on here? STP shows no blocked ports at the 5010 level. STP doesn't even show active for our FEX interfaces which I assume is inherited through the use of the "switchport mode fex-fabric" command.
Has anyone else seen this issue? or is this simply expected behavior?