HA Resetting VM Due to Tools Failure

Products

• VMware vCenter Server 8.0
• VMware vSphere ESX 8.x

Issue/Introduction

A virtual machine running within a vSphere cluster, particularly in a VMware Cloud Foundation (VCF) environment, unexpectedly resets. The event log shows the reset was initiated by vSphere HA with the reason: "VMware Tools heartbeat failure."

Error: This virtual machine reset by vSphere HA. Reason: VMware Tools heartbeat failure.

Reviewing the host logs indicates no signs of a Blue Screen of Death (BSOD) or a guest operating system failure. The Windows Event Log only shows an unexpected reboot.

Environment

 

• vSphere HA clusters.

• vCenter Environments deployed using VMware Cloud Foundation (VCF) 5.2

 

Cause

The unexpected VM resets occur when the vSphere HA advanced cluster setting das.iostatsInterval is configured to a value that is too low, specifically 0.

1. VCF Deployment: In certain VCF deployment versions, this setting is automatically configured to 0 to allegedly deactivate monitoring for management appliance I/O.

2. vSphere HA Behavior: A value of 0 or any value less than 50 is not recommended for VM Monitoring sensitivity. Setting it to 0 can cause vSphere HA to become overly sensitive or misinterpret normal resource fluctuations (such as storage or network I/O activity spikes) as a VMware Tools heartbeat failure, leading to an unnecessary and disruptive VM reset.

Resolution

The issue is resolved by resetting the das.iostatsInterval advanced cluster option to its recommended default value of 120.

1. Navigate to Cluster Settings: In the vSphere Client, go to your cluster, then select Configure > vSphere Availability > Edit > Advanced Options.

2. Add/Modify Option: Click Add and enter the following details, or modify the existing entry:

   • Option: das.iostatsInterval

   • Value: 120

3. Confirm: Click OK to save the changes.

4. Reconfigure: vSphere HA will reconfigure on the hosts. The cluster should no longer experience unexpected VM resets due to this configuration conflict.

Additional Information

The vSphere HA default value for das.iostatsInterval is 120 seconds. While vSphere allows setting this value to 0 (to deactivate the check), the official documentation advises against using values less than 50 due to the risk of unexpected VM resets. This issue is being addressed by product engineering to ensure VCF deployments do not incorrectly set this value in the future.