VM MAC showing up on two trunks and two VLANs

Patrick McHenry · ‎05-02-2011

Hi,

When finding a location of a VM I noticed that the MAC of the device is showing up on two trunks - the trunk that connects the access switch to the core (gi0/25) and the trunk that connects to the ESX host (gi0/17). The MAC address that is showing up on the trunk to the core is on VLAN 1 and the MAC on the trunk to the ESX host is on VLAN 17.

SWITCH#sh mac-address-table addr XXXX.XXXX.XXXX

Mac Address Table
-------------------------------------------

Vlan    Mac Address       Type        Ports
----    -----------       --------    -----
   1    XXXX.XXXX.XXXX     DYNAMIC     Gi0/25
17    XXXX.XXXX.XXXX    DYNAMIC     Gi0/17

The native VLAN of g0/25 is VLAN 1 and the native VLAN of int g0/17 is 4094 - not sure if this is a good idea?

Two questions:

It makes sense that the MAC associated with gi0/17 is on VLAN 17 as the VM is on VLAN 17, but why does it show up on the trunk to the core on VLAN 1?

Also, normally when I try to find a location of a device on the network (usually a fat client), the MAC doesn't show up on the trunk as well as the access-port? The only time it shows up on a trunk is when it is not connected to that switch on an access-port. Although in the the example of the post it is not on any access port so, maybe that is why?

Thanks for any thoughts, Pat.

Yogesh Ramdoss · ‎05-06-2011

Patrick.

When a host resides ONLY in vlan 17, then we should see that mac-address learnt ONLY in vlan 17, with no exceptions.

If the switch is learning x.x.x in vlan 1, it means that the traffic from VM has got bridged to vlan 1 somehow.

Few suggestions:

- Check on all your device to make sure none of them has "vlan dot1q tag native" in the global config. This command makes the dot1q to tag traffic even on native vlans. If one device has this enabled and not others, we may see similar issue.

- Check if any of the trunks has mismatching native vlans. e.g., on one end it is vlan 1 and other end it is vlan 100.

- Catalyst 2K/3K platforms support only 128 instances of spanning-tree. When we allow all vlans on the trunk (of course, after adding it to vlan database) and if it exceeds 128 instances, STP will be disabled on the vlans once we hit 128 limit. As a result, we may see similar issue.

http://www.cisco.com/en/US/docs/switches/blades/3120/software/release/12.2_55_se/configuration/guide/swstp.html#wp1170819

Regards,

Yogesh