Hello, I'm having some issues with separating traffic on my UCS. I have a Lab environment and a Prod environment. I'd like to have two servers in the UCS chassis be on the Lab net and 2 servers be on the Prod net. Both environments have the same network vlan architecture, so the vlans use the same vlan id's.
We assigned separate physical interfaces on the FI's for Lab and Prod (Fabric A port 21 Lab and Port 23 Prod, Fabric B port 21 Lab port 23 Prod), created PIN groups, one for Prod, and one for Lab, and assigned those groups to each port accordingly, both to the ethernet interfaces and the vnics on the service profiles. Also created separate vlan groups (Prod and Lab), and assigned the vlans to the specific ethernet uplink ports in the vlan group. (granted, they're the same vlans, since both environments are the same, but we assigned the proper ethernet ports for each environment)
We thought that was enough to keep the two environments separate, but I'm still seeing some traffic spill over between them. I can run a constant ping to one of the UCS servers in Prod, and every now and then I'll get a few timeouts, then will start replying again, which causes the application I'm running on those servers to disconnect frequently. When I disable (or unplug) the Lab interfaces on the FI, I never get the dropped pings and my application never has any problems with disconnects. This is telling me that there are times the UCS is directing traffic to the Lab interfaces like a load balancing or round robin thing. (The pings time out because I don't have that specific IP address in the Lab to reply from).
I'd like to know what else we might be missing on our configuration or what we're doing wrong. Is it a problem to have the same vlan id's in both environments? I would think that would be ok, but I'm not a UCS expert. Thanks for any insight or help someone can give me!
