ESXi host shows Not Responding in vCenter due to physical network ARP drops
Article ID: 452863
Updated On:
Products
Issue/Introduction
ESXi hosts may appear as "Not Responding" or "Disconnected" in vCenter Server when management traffic is interrupted at the physical network layer. This specific condition occurs when the ESXi host successfully responds to Address Resolution Protocol (ARP) requests from the default gateway, but those responses are dropped or not processed by the physical switch or gateway, leading to a loss of L3 connectivity.
Symptoms
- ESXi host shows Not Responding or Disconnected in vSphere Client.
- Host is unpingable from the default gateway.
- Management agents (
hostd,vpxa) are running but cannot communicate with vCenter. - Packet captures on the ESXi management uplink (
vmnicX) show:- Incoming ARP requests from the gateway IP.
- Outgoing ARP replies from the ESXi host.
- Repeated ARP requests from the gateway for the same host IP, indicating the previous reply was not received or honored.
Environment
- VMware vSphere ESXi 7.x / 8.x / 9.x
- Physical Network Environment
Cause
The physical network infrastructure (switches or routers) is failing to update its ARP table or is dropping ARP reply packets originating from the ESXi host. This prevents the default gateway from establishing a valid MAC-to-IP mapping for the host, breaking all routed management traffic (including vCenter heartbeats).
Resolution
To resolve this issue, perform the following steps to isolate and fix the physical network drop:
- Log in to the ESXi Shell or DCUI.
- Identify the management uplink using the command:
esxcli network nic list - Perform a packet capture on the uplink of the ESXi host to confirm if ARP replies are leaving the host.
- Review the capture. If the host is sending "ARP Reply" but the gateway continues to send "ARP Request", the issue is external to the ESXi host.
- Engage the physical networking team to:
- Inspect physical switch ports for errors or discards.
- Verify ARP timeout settings on the gateway.
- Check for MAC learning limits or security features (e.g., Dynamic ARP Inspection) that might be blocking the host's MAC address.
- To temporarily restore connectivity, try failing over the management traffic to a different physical uplink if available:
esxcli network nic down -n vmnic#### - Verify if the host reconnects via the redundant uplink.
Additional Information
Feedback
