Last week I setup PI 2.0 on an ESX server using the OVA template. I can access the PI through the WebGUI without issue.
Because of our network topology, I need to use 3 interfaces to properly manage our environment. Through vCenter, I added 2 additional interfaces, and rebooted the server. All three interfaces show up properly in the CLI as well as in the GUI. I configured all three itnerfaces with the appripriate IP addresses, and configured a single default gateway out interface Gi 0.
Gi 1 connects to our TACACs server as well as our swith infrastructure. Gi 2 connects to our WLAN infrastrcuture, and will be used to manage WISM2s (and the WLAN controllers in the new 4500 SUP-8 when the software gets released).
I managed to setup TACACS to our old ACS 4.2 server thgough Gi 1, as well as add a couple of switches to PI.
I could not however add the WISM2 through int Gi 2 (PI and controller in the same subnet). I could ping the controller from the PI, and viceversa, but could not poll the device, not access it via SSH.
After hours of troubleshooting and getting now where, I swapped interface Gi 1 and Gi 2, and reconfigured the IP addresses appropriatly, i.e. Gi 1 is now conencted to the WLAN infrastrcture, and Gi 2 connected to the switch infrastrcture.
After the change I could no longer access the PI using TACACS, nor could I update or add any devices connected through Gi 2. They still show as reachable (as I beleive the reachability is ICMP-based), but that's it. I had no issues connecting the WISM2 to the PI once the WLAN mgmt interface was moved to Gi 1.
Does anyone know of any restrictions in PI regarding the number of supported interfaces? I search bug tracker and the documentaiton, and could nto find anything related.
My only conclusion is that PI 2.0 does not properly support more than 2 interfaces.
Has anyone else run into this "feature"? Any advice would be appreciated.