%sw_matm-4-macflap_notif:

Bosefus

Earlier today we had several of these messages popping up and crashing one of our switches (3750) which I will call main switch for purposes of this post.

%SW_MATM-4-MACFLAP_NOTIF: Host XXXX.XXXX.XXXX on Vlan XXX is flapping between port g1/0/2 and g1/0/13

We had several sets of these messages for different pairs of ports.

The setup was a lab switch (HP) which and the servers connecting into it. This lab switch then connected into our main switch (3750) via 1 Ethernet cable.

We unplugged the lab switch that a server which had a load balancing plugged into it.

I read this post which states what the problem was or could have been. Which would be the fact that the same mac was coming up on two different ports causing the messages.

http://www.********.com/forum/index.php?showtopic=144445

My question is, why would this affect our main switch? The flapping should only occur on the lab switch that the server is plugged into, not our main switch.

If the server was plugged into the main switch on two different ports I could understand the messages popping up on the main switch. However, since there was only one cable connecting the lab switch to our main switch I cannot understand why the flapping was occurring on the main switch.

Any ideas?

-Thanks

networker050184

Was there a cable from the lab switch to another switch? What is connected to port g1/0/2 and g1/0/13?

jason_lunde

So which port on "main" does the lab switch plug into? 1/0/2 or 1/0/13? After this...what is plugged into that other port?

Is that vlan in the message XXX connected any other way in the environment that would form a loop? Think about how a switch reads the mac addresses....by source...and troubleshoot from there.

damn...you beat me

Bosefus

The lab switch goes into a blade server that has a built in switch.

Lab switch was plugged into g1/0/2 and a end device (PC) is plugged into the other port.

There are other vlans on the main switch but the HP switch is a dumb switch plugging into a port on the main switch which is in a vlan. The HP itself does not have any vlans configured.

So this is or could be some sort of looping issue?

networker050184

You are probably looping through the blade center switch. What kind of module do you have in there? If its a Cisco module it should be pretty easy to find with your normal troubleshooting. If they are the Nortel models it may be a bit more trouble to track down.

jason_lunde

Ya I agree with networker...in these cases I try to figure out what exactly the mac address belongs too. If you dont know..then use the show arp | include XXXX (where xxxx is the last four of the mac address) in conjunction with show mac-address-table. This should give you the ip and vlan, and port connectivity for that mac address.... You should be able to derive the rest from there. Find the culprit, check its connectivity, and you should be golden.

Bosefus

Cool thanks for the replies.

I thought it was something with the blade server.

I have never dealt with them before, and I did not see how it could be a dumb switch or the 3750.

Will see if there is a management interface for the switch in the blade server.

According to the manual the switch in the blade chassis switch is multilayer.

Also, if the blade center switch is looping, how would that cause a mac address port change on the 3750?

The connection is 3750, HP dumb switch, blade center.

Thanks for all the help and advice.