vSphere HA behavior and DRS contention during frequent NIC flapping in vSAN clusters

Products

• VMware vSphere ESXi

Issue/Introduction

• ESXi host reports Host isolated is true followed immediately by Host isolated is false in fdm.log.
• Virtual Machines (VMs) appear disconnected or orphaned in vCenter Server.
• Cluster experiences heavy CPU contention and excessive DRS migrations.
• vSphere HA does not trigger VM failover to other hosts despite connectivity loss.
• vSAN-enabled clusters show isolation events triggered by the storage network.

Environment

• VMware ESXi 8.x
• vSphere HA
• vSAN Cluster

Cause

This behavior is expected when a network interface (vmnic) flaps frequently. vSphere HA requires a sustained isolation period (default 30 seconds) to initiate isolation actions and VM failovers. If the heartbeat is restored within this interval, the host exits the isolated state before failover triggers. Rapid state transitions cause DRS to perceive cluster instability, leading to continuous VM migrations and CPU overhead.

Resolution

1. Inspect physical switch ports and cables for the ESXi host for link flaps or CRC errors.
2. Review switch logs for spanning-tree (STP) events or port-security triggers occurring in regular increments.
3. Ensure network redundancy for management and vSAN traffic to prevent single-link failures from isolating the host.