Host not responding temporarily causing vSphere HA to failover VMs

Products

• VMware vSphere ESXi

Issue/Introduction

• ESXi host goes into not responding state for 2-5 minutes
• While in not responding state, VMs failover to other hosts
• /var/log/fdm.log shows Election State Change Slave => Startup : Lost master
• Either /var/log/vmkernel.log or /var/log/vmkwarning.log shows:
  YYYY-MM-DDTHH:MM:SS Wa(180) vmkwarning: cpu0:2656###)WARNING: Heartbeat: 632: PCPU# didn't perform a heartbeat check for ### seconds

Environment

ESXi 8.x

Cause

The physical CPU freezes and doesn't heartbeat for 2-5 minutes. Thus, the ESXi host cannot carry out CPU operations and locks up.

During this time, the host will not log, VMs will fail over and the host will be unaccessible.

Resolution

Engage with your hardware vendor to investigate the CPU referenced in the log message.