Network connectivity loss to a VM after vMotion or cross-vCenter migration
Article ID: 426004
Updated On:
Products
Issue/Introduction
- Continuous ping to the VM IP address fails as soon as a vMotion or cross-vCenter migration completes.
- Connectivity is eventually restored after manually disconnecting and reconnecting the VM’s virtual network adapter.
- Packet captures on the host uplink confirm that Reverse ARP (RARP) and standard ARP packets are successfully egressing the ESXi host.
- The "Notify Switch" option is correctly enabled on the Distributed Switch (vDS) or Standard Switch.
Environment
VMware vSphere ESXi
Cause
When a VM moves to a new host, the physical switch port connected to the destination host must learn the VM's MAC address to update its forwarding table (MAC table). This is typically triggered by a Reverse ARP (RARP) packet sent by the ESXi host on behalf of the VM.
The connectivity failure occurs when:
The RARP packet or subsequent ARP requests from the VM are disregarded or dropped by the physical network infrastructure.
There is a delayed MAC table update on the physical switches.
Resolution
Engage the physical networking team to investigate delayed MAC learning or port-security settings on the upstream switches and Verify that the physical switch ports are configured for the correct VLANs and are not in a "Discard" or "Blocking" state for the VM's MAC address.
As a workaround, logging into the Guest OS and initiating a continuous ping to the default gateway can force the physical switch to learn the new MAC location.
For more information on IP to MAC mapping and "Notify Switch" settings, see IP to MAC mapping, GARP, RARP and Notify Switch setting for Virtual Machine Connectivity
Additional Information
For related information, check the KBs below:
Feedback
