Customer has hit a minor issue which he has been fiddling with for a few days and came to the conclusion that it may be some kind of compatibility issue with their HP Apollo servers and ACI.
We are running LACP in vPCs down to the Apollo servers but only 1 interface can forward at any one time. The other interface goes into a hot-standby state and will only forward if we shut down the active interface. This behaviour is the same for all 8 ESXi hosts we have running in our pilot. We have other LACP vPCs to our Solidfire storage nodes and not seeing the same issue with them. We've checked the max links allowed in the portchannel and these at set to the default of 16 on the ACI side, and 2 in vSphere which both should conform and not place the interfaces into hotstandby.
As a test I configured a standard LACP port-channel on one of the NXOS N9Ks and pathed one of the ESXi hosts into the NXOS N9K and the portchannel came up perfectly fine - no hot-standby and no configuration changes on the ESXi host.
I then patched the NXOS N9K into the ACI vPC (in place of an ESXi host) and this came up fine too, both interfaces were forwarding and no configuration changes in ACI.
We are using VXLAN over LACP.
