stuck mac address

[dennislarge]

This has me stumped. 6248UP FIs, 5108 chassis with 3 B200M4 blades running recent ESXi, 3.1(1g) firmware thruout.

One of the blades has mac stuck on at least one of the mgmt vnics. The stuck mac is a valid one in the pool, and was assigned at one point to this server, but that profile has been deleted and the mac shows as unassigned. The server is on different template, and it's assigned mac is what I see on the esx console. But the old one still shows up in the FIs. I've cleared the dynamic mac pool on both FIs, forced a failover, decommissioned and reassociated the blade. Have not actually restarted the FIs, which I could easily enough but would prefer not to. (No production in place at the moment, but would like to resolve as if it was.)

Any ideas short of a tac case?

[Kirk J]

Greetings.

What 'Veth' interface is associated with that learned MAC?

Do you have these built out as boot from san?  When ESXi is first installed the VMK0 inherits it's MAC from the physical NIC.  If that instance of ESXi is present (abeit running from another blade/service profile), then you will still see that mac in use.

I'm assuming the mac aging time is being updated?

If this is the scenario, see the workaround at https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1031111

Thanks,

Kirk...

[dennislarge]

No boot from san. played with the mac aging, no help. Ended up assigning different vnics, saw the new ones pop up, then reverted and the right ones showed up. Sounds like maybe the vswitch was holding on to it. So weird.

Thanks for the suggestions. Gives me some more commands to add to the pile. A bigger toolbox is always a good thing.