Cisco Support Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Announcements

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. And see here for current known issues.

New Member

Ethernet address on two trunked switches at the same time

I have two access layer switches sharing a dot1q trunk ( a port channel with four members), and an HP blade server with one connection (a port channel with two members) into each..  The blade server's connections into each switch is also a dot1q trunk tagging just one vlan.  In each connection from the blade server to each switch the mac addresses look very much alike.  I determine this by writing a script to log into both switches at once and run the "show mac add int" command on each port channel to the blade server.  In tcpdumps the commands are sent typically about .1 - .4 seconds apart on average.

So if the "show mac addr int" command is being run .1 - .4 seconds apart on each switch and have a number of mac addresses in common, would this indicate that these ethernet addresses are flapping in a way that might cause interruptions in packet delivery?  Some are experiencing missed pings.  No VPCs are used in this configuration -  just a pair of either 6509 or 7018 switches connected via a dot1q port channel.

Everyone's tags (2)
4 REPLIES
Bronze

I'm going to have see some

I'm going to have see some topo drawling and configs for that one...

- Be sure to rate all helpful posts
New Member

Here is a basic pic of how

Here is a basic pic of how this blade server is connected to the two switches (nexus 7018s):

 

The question in my mind is if the commands to see the mac table of each respective port channel on each respective switch are issued within one - five tenths of a second apart, is that sufficient to show that the mac address is truly on both switches at the same time?  We don't have VPCs set up, and there are anywhere from one - 2 dozen duplicate mac addresses like the one above at any given time the two mac address tables are checked.  I suspect the blade server is attempting to somehow team or channel in some way, and am going to a meeting with the sysadmins and vendor to try and find out why it's doing this.

HelloThe cause of the

Hello

The cause of the flapping links could be a mixture of a couple of things:

Link/port aggregation protocol being used ( i think nexus only support LACP)
Stp negotiation
Port channel load balancing defind

 

Is it possible you can post your interface and port-channel config and load balancing mode.

sh run int x/x
sh run int port-channel x/x
sh port-channel load-balance

 

res

Paul
 

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

Are you using the appropriate

Are you using the appropriate NIC teaming on the HP blade server?  If you setup a port channel on the switch side, use LACP on those switches and on the HP server use its NIC teaming software with the 802.3ad setting.

99
Views
0
Helpful
4
Replies