Unable to decrease "Host failures cluster tolerates" value in vSphere HA Admission Control

Products

• VMware vCenter Server

Issue/Introduction

When attempting to reduce the number of "Host failures cluster tolerates" in the vSphere HA Admission Control settings, the value automatically reverts to a higher/original number immediately after clicking "OK" or "Save".

• The vSphere Client UI does not display an error or task failure.

• The settings appear to save successfully but return to the original higher value upon refreshing or re-opening the configuration.

Environment

vSphere 8.x

Cause

This behavior occurs when the existing Reserved Failover Capacity (CPU and Memory) is too high to support a lower host failure tolerance setting.

vSphere HA Admission Control calculates the resources required to satisfy the failover policy. If internal reservations (such as those created by VM/Host Group rules, high VM reservations, or strict Admission Control policies) exceed the available unreserved resources, the cluster enforces a minimum "Host failures cluster tolerates" value to ensure the environment remains compliant with its resource availability.

In this scenario, the cluster prevents the user from lowering the failure tolerance because the current resource overhead/reservations already effectively "consume" the capacity of multiple hosts.

Resolution

To resolve this issue, the reserved capacity must be reduced within the cluster to allow Admission Control to accept a lower host failure tolerance.

1. Check Current Capacity: Navigate to the Cluster > Monitor > vSphere HA summary page and check the Reserved failover CPU/Memory capacity. If these values are high (e.g., above 50%), identify what is consuming that reservation.

2. Review Affinity/Anti-Affinity Rules: Check for VM/Host Group rules. These rules can force high resource reservations on specific subsets of hosts, driving up the failover capacity requirements.

3. Modify or Remove Rules: Disable or remove unnecessary Host Group rules or VM affinity rules that contribute to high resource overhead.

4. Adjust Admission Control: Once the Reserved Failover Capacity has decreased (e.g., to 15–20% range), attempt to change the "Host failures cluster tolerates" value again. The setting should now save and persist as expected.

5. Verify VM Reservations: If the issue persists, check if individual Virtual Machines have high CPU or Memory reservations, as these also contribute to the total capacity required by the HA calculation.