DRS Migrating All Virtual Machines to a Single Host Due to HA Admission Control Constraints

Products

• VMware vCenter Server

Issue/Introduction

• VMware vSphere Distributed Resource Scheduler (DRS) is enabled and set to Fully Automated mode.
  
  
• Most Virtual Machines are migrated to a specific ESXi host in the cluster. This behavior is observed when any host is placed into maintenance mode.
  
  
• DRS does not distribute workloads across all available hosts in the cluster despite having sufficient aggregate resources.
  
  
• The task to enter maintenance mode proceeds, but all Virtual Machines move to the same destination host.
  
  
• There are very few or no CPU and Memory reservations configured for the Virtual Machines in this cluster.

Environment

VMware vCenter Server 8.x

Cause

It occurs due to restrictive vSphere HA Admission Control settings. When Admission Control is configured with a high failover capacity percentage, it reserves a significant portion of CPU and Memory resources across the cluster to ensure availability during a host failure.

During a DRS migration event (such as placing a host in maintenance mode), DRS must select a destination host that satisfies two conditions simultaneously:

• The host must have enough unreserved physical resources for the Virtual Machine.
  
  
• The migration must not violate the HA Admission Control policy.

If only one host in the cluster meets the strict resource requirements enforced by the current Admission Control policy, DRS is forced to move all Virtual Machines to that specific host, resulting in an uneven distribution of workloads across the cluster.

Resolution

Adjust the HA Admission Control settings to reduce the reserved failover capacity, allowing DRS more flexibility in host selection.

Procedure:

1. Log in to the vSphere Client.
   
   
2. Select the affected Cluster in the inventory.
   
   
3. Navigate to Configure > vSphere Availability.
   
   
4. Click Edit next to vSphere HA.
   
   
5. Select the Admission Control tab.
   
   
6. Review the Define host failover capacity by setting. If it is set to Cluster Resource Percentage, note the current values for CPU and Memory.
   
   
7. Reduce the percentage to a value that meets the specific cluster redundancy requirements (e.g., 8% or 10%).
   
   
8. Click OK to save the changes.

Verify the resolution by manually triggering a vMotion or placing a host in maintenance mode to confirm Virtual Machines are now distributed across multiple hosts.