cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
994
Views
0
Helpful
2
Replies

Nexus 7000 and Dual-Homed FEX issue

dcolladoe
Level 1
Level 1

Hi,

I have two Cisco Nexus 7009 with N7K-SUP2 connected in vPC through N7K-F312FQ-25 cards. I also have 5 FEX N2K-C2348TQ-10GE connected in dual home, the 7K have the IOS 7.2(1)D1(1) that allows this topology.

The vPC domain is configured as follows and all vlans are in the peer-link:

vpc domain 1

peer-switch
role priority "X"
peer-keepalive destination a.a.a.a source b.b.b.b
no auto-recovery
ip arp synchronize

I have one server connected to one FEX, port Eth101/1/1 (in example), and the issue is that I can't ping it:

- Server is in the same vlan than interface vlan on Nexus 7K.

- Server is known from N7k via MAC or ARP.

If I shutdown the link between N7K-1 and FEX, N7K-2 starts to ping server, not N7K-1, but N7K-1 see the mac of the server through peer-link. With "no shutdown" the server is not reachable again from both N7Ks. Now, if I shutdown the link between N7K-2 and FEX, the N7K-1 starts to ping server, and also N7K-2 (that see the mac in the peer-link). When I "no shutdown" the interface, traffic is nice again ans server is reachable from both N7Ks. If I restart the server, the problem appears again.

Do you think that could be any bug regarding dual homed FEX?.

Thank you in advance.

2 Replies 2

Rajeshkumar Gatti
Cisco Employee
Cisco Employee

I suspect that ARP is not getting across when there is ping failure. You can check the ARP table on both the 7k and the host to confirm if the ARP entry for the corresponding broken IP exist. I recommend opening a TAC case as it will help use some of the internal capture tools on the 7k to check the flow of ARP and IP packets in a broken state.

-Raj

Hi Raj!

Seems to be ARP, but both N7Ks have an entry for the IP address of the server.

Regards,

David.

Review Cisco Networking for a $25 gift card