mac-flap issue

[manassarkar89]

Hi,

I have 16 switches and 3 servers in my network.In server room,i have one cisco WS-C2960S-48TS-L and one WS-C2960S-24TS-L connected to each other.Two eth ports of each server are being timmed and connected to the 2 switches.Every thing is fine but i am getting mac-flap log in both switches.can anyone tell me how can i stop this?

48TS-L is the root bridge of the network and there is only 1 vlan in the network.all the ports are in same vlan of both switches.

here is a sample of the logs of both siwitches :-

*Feb 1 08:24:28.382: %SW_MATM-4-MACFLAP_NOTIF: Host 0015.5d07.1903 in vlan 101 is flapping between port Gi1/0/48 and port Gi1/0/3

*Feb 1 08:26:34.992: %SW_MATM-4-MACFLAP_NOTIF: Host 0015.5d07.1e01 in vlan 101 is flapping between port Gi1/0/48 and port Gi1/0/6

*Feb 1 08:25:40.351: %SW_MATM-4-MACFLAP_NOTIF: Host 0015.5d07.1e05 in vlan 101 is flapping between port Gi1/0/24 and port Gi1/0/15

Thanks in advance.

[Richard Primm]

Hello Manas,

We see this often and the resolve is almost always in the configuration of the server side nics or virtual nics.  I would read up on the vendor documentation and verify their recommendation for deployment. They will usually also give examples of their preferred method of configuration on Cisco switches.

HTH

Luke

[devils_advocate]

As Richard has suggested, this is often caused by a misconfiguration at the server side.

Effectively the switch is recieving frames on two seperate ports which have the same Source MAC Address. If teaming is setup correctly, the switch will be able to handle this but if not, you will end up with MAC Flapping.

[paul driver]

Hello

It occurs when you have a loop between the switches relating to ports bypassing the stp process and enter forwarding state either due stp being turned off or having portfast enabled.

when you have a server connectiing itself to two switches on portfast enabled ports, this can cause this type of issue you are seeing- Check if this is the case and disable the portfast feature on those ports

For trunk ports or if port aggregation check the link negotiation type (usually LACP for servers)  is applied the same on both sides of the etherchannel and teamed server nics

res

Paul

Please don't forget to rate any posts that have been helpful.

Thanks.

[manassarkar89]

Hi paul,

thanks for the kind reply.

first of all portfast is not enabled on the ports and one of switch is root bridge of the network,so stp is on.

and also we have only one vlan in the whole network which is vlan 101.so there is no trunk ports.

but i have one question,two ports are of two different switches,can we make an etherchannel out of it???

thanks in advance.

manas

[acampbell]

Manas,

If you physically connect the NICs from your servers to sepeate switches you will

need to change the teaming set up on the servers to "Network Fault Tolerant" working.

If you want server Teaming to be NIC load balancing you will need the physical connections from

the NICs to connect to the same switch. Then use LACP between the switch and the servers.

Regards,
Alex.
Please rate useful posts.

[dinesh.thathanath]

Interesting.

I see the MAC flaps on my 3850s when a user keeps moving around from one AP to another. Logically in my case it makes total sense as I use local switching for my Wireless setup and my WLCs reside at a remote location.

I have not seen similar logs on 3750-x for the exact same setup.

Dinesh

Sent from Cisco Technical Support iPhone App