vSphere HA configuration fails with host state stuck in Election
Article ID: 438412
Updated On:
Products
Issue/Introduction
vSphere HA configuration on the cluster remains in a pending state, and the ESXi hosts report that their HA state changed to "Election".
Disabling and re-enabling HA on the cluster does not resolve the behavior. When reviewing the Fault Domain Manager (FDM) logs on the ESXi hosts, errors are present indicating the host is unable to reach the HA master.
Standard ping tests and TCP/UDP port 8182 communication may succeed, but ping tests utilizing an MTU payload of 8972 fail to reach the destination
Environment
VMware vSphere Esxi,
VMware vCenter Server
Resolution
The issue is caused by a Maximum Transmission Unit (MTU) configuration mismatch between the VMkernel adapter and the virtual switch.
The VMkernel port is configured for jumbo frames (MTU 9000), but the underlying vSphere Standard Switch (VSS) is configured with the default MTU of 1500, causing HA packets that exceed 1500 bytes to be dropped.
Additional Information
Connect to the vCenter Server using the vSphere Client.
Select the affected ESXi host and navigate to Configure > Networking > Virtual switches.
Identify the vSphere Standard Switch (VSS) or Distributed Virtual Switch (DVS) hosting the VMkernel adapter used for HA network traffic.
Edit the virtual switch settings and change the MTU from 1500 to 9000 to match the VMkernel adapter.
SSH into the ESXi host and validate end-to-end jumbo frame connectivity using
vmkping:vmkping -I vmk0 -d -s 8972 <REDACTED_IPS>Once the jumbo frame ping is successful, return to the vCenter Server.
Right-click the affected cluster and select Reconfigure for vSphere HA, or toggle HA off and back on to allow the hosts to successfully complete the election process.
Feedback
