Dropping received frames from duplicate VSM saddr

[bjoern.strassburg89]

Hello,

today I installed the Nexus 1000v on three ESX servers, I think everything is working fine but there is a Log Message which occurs every two minutes. The  following Message will be displayed: "%KERN-1-SYSTEM_MSG: Dropping received frames from duplicate VSM saddr (0x1010000) - kernel"

Some details of the servers:

Two of them are connected with six NICs to a Cisco 3750 Stack. I've configured two portchannels on the Nexus 1000v one with 4 NICs (data uplink) the other with 2 NICs (managment)

One of them is connected to a Cisco 3750 and Cisco 4006 I've also configured two portchannels on the Nexus 1000v one with 4 NICs (data uplink) the other with 2 NICs (managment) here with vPC-HM

Does anybody know something about these log message, is there a fault with the Portchannels? How should it be with portchannels on the upstream switches?

[abbharga]

Hi,

What you are seeing is basically deuplication of heartbeats between the VSM and VEM. This genreally happens when mutiple interfaces carrying the control vlan are not bundled together either on the VEM or on the upstream switch, casuing the packets to be duplicated.

On the upstream switch all the interfaces coming from the same port-channel from the same VEM should be bundled together, in short if you have more then one interface from a particualr port-channel on the VEM on the same switch they should be bundled together.

In you case for the two servers with 6 NICs going to the same 3750 stack, you need to configure 4 POs on the 3750 stack. One per server for the 4 data uplink interfaces and 1 per server for the 2 management interface.

For the third server going to 3750 and the 4006 switch, if there are more then one interface on each switch for the data uplink bundle them together as for the management uplink if there is only one uplink per switch no need to bundle it.

I hope this answers your query!

./Abhinav