Managing vSphere HA and DRS settings to prevent host isolation restarts during network maintenance
search cancel

Managing vSphere HA and DRS settings to prevent host isolation restarts during network maintenance

book

Article ID: 448311

calendar_today

Updated On:

Products

VMware vSphere ESXi VMware vCenter Server

Issue/Introduction

In VMware vSphere environments, planned network infrastructure upgrades (such as core switch replacements) can result in a total loss of management and VM network connectivity for ESXi hosts. If the hosts and virtual machines (VMs) are left powered on while the network is down, the vSphere High Availability (HA) cluster may incorrectly identify this as a host failure or isolation event.

Symptoms:

  • vSphere HA triggers a "Host Isolation" or "Host Dead" event.

  • Virtual machines are unexpectedly powered off or restarted on other hosts.

  • DRS migration tasks fail due to a lack of network connectivity for vMotion.

  • vCenter Server alarms for Host connection failure or Host is not responding.

Environment

  • Product: VMware vSphere ESXi

  • Versions: 7.x, 8.x (including 8.0 Update 3)

  • Component: High Availability (HA), Distributed Resource Scheduler (DRS)

  • Configuration: Environments with shared storage (SAN/NFS) where hosts remain powered on during network maintenance.

Cause

By default, vSphere HA uses network heartbeats to monitor host liveness. If a host loses management network connectivity for more than 30 seconds, the HA Master host declares that host as "Dead" or "Isolated." If the Host Isolation Response is configured to restart VMs, HA will attempt to move those workloads. During a cluster-wide network outage, these migration and restart attempts will fail, leading to unnecessary downtime and potential data inconsistency since the underlying storage remains active.

Resolution

To prevent unintended cluster responses during scheduled network maintenance, follow these steps:

1. Disable vSphere HA Host Monitoring

This prevents HA from reacting to the loss of network heartbeats.

  1. Log in to the vSphere Client.

  2. Navigate to the Cluster in the inventory.

  3. Click the Configure tab.

  4. Under Services, select vSphere Availability.

  5. Click Edit.

  6. Uncheck the Enable Host Monitoring checkbox.

  7. Click OK.

2. Set DRS to Manual Mode

This ensures that no vMotion migrations are initiated while the network is unavailable.

  1. Navigate to the Cluster > Configure tab.

  2. Under Services, select vSphere DRS.

  3. Click Edit.

  4. Change the Automation Level to Manual.

  5. Click OK.

3. Post-Maintenance Restoration

Once the network maintenance is complete and physical connectivity/routing has been verified:

  1. Navigate back to vSphere Availability and re-check Enable Host Monitoring.

  2. Navigate back to vSphere DRS and return the Automation Level to its previous state (e.g., Fully Automated).

Additional Information

Powered by Wolken Software