When I use DCNM to discover a Nexus 5548, entering its mgmt0 address (172.16.x.190) as the seed, DCNM finds it but lists it as 188.8.131.52. It takes a long time to discover for some reason, too. When I open the fabric, it cannot access the switch until I change its address manually, at which point it seems to fully access the device. However, the fabric seed address is still listed as 184.108.40.206.
Any idea what's going on here?
To the discovery process works properly you must config the ALL parameters for authentication, ssh and snmp in the N5K.
In your case I think the following commands are very important:
#snmp-server enable traps
#snmp-server source-interface mgmt 0
I have this problem too.
I'm adding the seed address as the vlan management address, however, after an initial dialog it then always tries to talk to the management interface - which i've set up as private vPC keepalive link (220.127.116.11 in my case).
I guess its picking this up from the FCS database, but I can't find anywhere how to change it;
nexus# sh fcs database
FCS Local Database in VSAN: 1
Switch WWN : 20:01:01:01:01:01:01:01
Switch Domain Id : 0x67(103)
Switch Mgmt-Addresses : snmp://18.104.22.168/eth-ip
Fabric-Name : 20:01:01:01:01:01:01:01
Switch Logical-Name : nexus
Switch Information List : [Cisco Systems, Inc.*N5K-C5596UP*5.2(1)N1(7)*20:00:00:00:00:00:00:00]
In my case, the problem ended up being that the customer had unknowingly configured the management interface with 1.1.1.x. Since this wasn't a routable subnet on their network, DCNM just continued to use the discovery address as its point of control, but it was always called 1.1.1.x in the list.