It gets complicated. The way this works is that IBM designs and controls the chassis and Cisco designs the Nexus 4000 switches for their chassis. But IBM still wants to retain control over the chassis via an advanced management module or AMM.i.e. IBM feels they should be able to control the Nexus 4000 module via the Advanced Management Module.
There are 2 management ports on the nexus 4000: one external 10/100/1000BASE-T port (mgmt0) and one internal port (full-duplex 100 Mbps) [mgmt1] connected to the AMM.
There are 3 different VRFs (show vrf all)
• Default – Typically used for “in-band” management (not via mgmt0 or mgmt1 but a layer 3 vlan interface)
• Management – Used for “out-of-band” management by interface mgmt0. This is the typical mgmt port and VRF you would find on the nexus 7000.
• Chassis-management – Used for “out-of-band” management by interface mgmt1 (via AMM web GUI or pass through)
Because network engineers are used to managing switches via the CLI, I typically see mgmt1 in "shutdown force" and a static ip address on mgmt0 for out of band management via the front external mgmt 0 port. But if you want to manage the module via the IBM AMM GUI, then you should use the mgmt1 and allow the AMM to give it an ip address via the IO module static ip address and shut down mgmt0. If you want both, then they need 2 different ip addresses.
I believe that the AMM will allow "pass through" mode to the Nexus 4000 CLI so you can reach the CLI via it.
Further information:
http://www.redbooks.ibm.com/abstracts/tips0754.html
http://www.cisco.com/en/US/docs/switches/datacenter/nexus4000/nexus4000_i/sw/configuration/guide/rel_4_1_2_E1_1/sol.html#wp1072280
On previous blade switches without VRFs, it was possible to loop packets through the AMM and back into the network when it was mis-configured since this internal interface is never blocking. The AMM guide might give you more information since it treats the switch as an IO module.
