Migration of Avi Controller and Service Engine VMs from one ESXi Cluster to another

Products

• VMware Avi Load Balancer

Issue/Introduction

• The Avi Controller and Service Engine (SE) virtual machines need to be migrated (vMotion) from one ESXi host cluster to another within the same vCenter environment, ensuring minimal traffic disruption and maintaining cluster integrity.

Environment

• VMware vCenter
• ESXi Clusters
• VMware Avi Load Balancer

Resolution

Controller Migration

1. Break the Existing Controller Cluster

   • Log in to Avi GUI → Administrator → Controller → Nodes

   • Remove both follower nodes from the cluster

   • All three nodes will now operate as standalone Controllers

   • Reference Document: Dismantling the Controller Cluster

2. vMotion Controller Nodes to the New ESXi Cluster

   • vMotion Follower-1 to the new ESXi host cluster

   • vMotion Follower-2 to the new ESXi host cluster

   • vMotion the Leader to the new ESXi host cluster

   • Verify all three Controller VMs are powered on and reachable

3. Rebuild the Controller Cluster

   • Ensure ports 22, 443, and 8443 are open between all three controller nodes for successful cluster formation

   • Re-join both follower nodes to the Leader from: Administrator → Controller → Nodes

   • Reference Document: Avi Load Balancer Controller Cluster

4. Validate Cluster Health

   • On the Leader controller, run:

     > show cluster status
     > show cluster nodes

---

Service Engine Migration

A. For Active/Standby SE Groups

1. Migrate the Standby SE

   • vMotion the Standby SE to the new ESXi cluster and verify that ports 22, 443, and 8443 are open between the Controller and the Service Engine for management communication.

   • Check the status of SE on leader controller GUI: Infrastructure→Cloud Resources→Service Engine

2. Switchover Virtual Services

   • Initiate switchover so the migrated SE becomes Active: Run the below command on leader node 

     > switchover serviceengine <active-se-name>

   • After running the switchover command please check the VS health status on controller GUI: Applications → Virtual Services

   • Traffic will now flow through the newly migrated SE. 

3. Migrate the Original Active SE

   • vMotion the previous Active SE to the new ESXi cluster.

   • This SE will now be standby SE.

4. Reconnect if Required

   • If the SE does not auto-connect after migration, run the below script from leader controller node: 

     Path: /opt/avi/scripts/ 

     # python3 connect_se.py --se ipaddress --username admin --password password

B. For Active/Active or N+M SE Groups

1. Migration can be performed in any order since traffic is already load-balanced.

2. vMotion SEs one by one to the new ESXi cluster.

3. No switchover is required because redundancy preserves service availability.

Note:

• Make sure that new ESXI Cluster have the correct portgroup access on ESXI host before power on the Controller and SE VM.

Additional Information

• It is recommended to set DRS to Manual or Disabled (VM Overrides) for Controllers VM. DRS rule or exception (VM Overrides) to exclude the Controllers VM. 
  Reference Article: DRS and Anti-affinity Rules