The steps in this section are for troubleshooting where packets are being lost for an overlay switch.

For a VLAN-backed logical switch, go to Troubleshoot Packet Loss for a VLAN logical Switch or When ARP Is Resolved.

Before performing the following troubleshooting steps, run the command arp -n on each VM. If ARP is successfully resolved on both VMs, you do not need to perform the steps in this section. Instead, go to the next section Troubleshoot Packet Loss for a VLAN logical Switch or When ARP Is Resolved.

Procedure

If both endpoints are ESXi and ARP proxy is enabled on the logical switch (only supported for overlay logical switches), check the ARP table on the CCP and the hypervisor.

On the CCP:

controller1> get logical-switch 5000 arp-table

On the hypervisor, start NSX CLI and run the following command:

host1> get logical-switch <logical-switch-UUID> arp-table

Fetching the ARP table only tells us whether we have the correct ARP proxy state. If the ARP response is not received via proxy, or if the host is KVM and does not support ARP proxy, the datapath should broadcast the ARP request. There might be a problem with BUM traffic forwarding. Try the following steps:

  • If the replication mode for the logical switch is MTEP, change the replication mode to SOURCE for the logical switch from the NSX Manager GUI. This might fix the issue and ping will start working.

  • Add static ARP entries and see if the rest of the datapath works.