DRS violates VM-Host Soft Affinity rule causing Network outage on the VM

Products

• VMware NSX
• VMware vCenter Server
• VMware vSAN

Issue/Introduction

Symptoms:

• vSAN Stretched Cluster with no stretched L2 Network between the sites
• Most VMs running on the "Preferred" site
• "Soft" or "Should" DRS VM-Host  Affinity Rule configured to protect against any site failures
• Multiple VM-Host Groups with Anti-Affinity rules configured in the same cluster
• When a host fails or enters Maintenance mode, DRS is seen violating the VM-Host Affinity rule and moves the VM to the "Secondary Site" Host, causing the VM to lose Network connectivity

Cause

• This is due to a low number of ESXi hosts in the VM-Host Anti-Affinity group than there are VMs.
• DRS requires that if there are N VMs with Anti-Affinity configured to each other, then you need a minimum of N + 1 hosts in the preferred host group to find placement within the group.
• For example, if you have 3 VMs in the group, you need 4 Hosts in the cluster for placing the VMs in the event of a failure or Maintenance.

Resolution

Add more hosts to the cluster or reduce the VMs in the Anti-Affinity group