Troubleshooting packet loss on Nexus 3000

[fwrusso]

Hello,
We are experiencing, on our VMware based data center infrastructure, a problem where sometimes, after a Vmotion, virtual machines lose network connectivity for a time period (around 30secs), sufficient to generate application errors.
I suspect that the problem arises from lost RARP packets - the packets generated by ESXi hosts after a vmotion, with the specific purpose of notifying switches about the new location of the VM.

How could I troubleshoot this issue, firstly in order to confirm the hypothesis above, and, in such case, to remove the cause of such packet loss?

Hosts are connected each to a couple of Nexus C3064PQ, on 10Gbps interfaces, as ToR switches, which are in turn connected to a couple of Nexus C3132Q-40GX as spine.

Many thanks for your advice

[M02@rt37]

Hello @fwrusso 

Ideas in order to troubleshoot this issue:

- Check the ARP cache on the virtual machine and on the switches to make sure that the correct IP-to-MAC address mappings are present. 

- Use a network traffic monitoring tool, such as Wireshark, to monitor network traffic during a vMotion operation. Look for any dropped RARP packets or other network errors that may indicate the cause of the connectivity loss.

- Make sure that the interfaces on the switches and the network are not oversubscribed. You can use the "show interfaces" command on the switches to check utilization levels.

- Check for spanning tree issues (if you are on a VPC Design) : Verify that spanning tree is configured correctly and that there are no loops in the network.