An NSX Controller cluster must have at least three members. If an NSX Controller appliance becomes inoperable or if you want to remove it from the cluster for any other reason, you must first add a new NSX Controller appliance to make a four-member cluster. Once the fourth member is added, you can remove an NSX Controller appliance from the cluster.

Prerequisites

  • Verify through troubleshooting that the appliances are not recoverable. For example, these steps may recover the appliances without having to replace them.

    • Verify that the appliances have network connectivity, and resolve if not.

    • Reboot the appliances.

  • Verify that you know the version of the NSX Controller that you are replacing and have an appropriate installation file (OVA, OVF, or QCOW2) of the same version available.

Procedure

  1. Install and configure a new NSX Controller.

    See the NSX-T Installation Guide for information and instructions about these steps.

    1. Install a new NSX Controller appliance.

      The version of the new NSX Controller must be the same as the NSX Controller it is replacing.

    2. Join the new NSX Controller with the management plane.
    3. Join the new NSX Controller with the control cluster.
  2. Shut down the NSX Controller you want to remove from the cluster.
  3. Log in to another NSX Controller and check that the NSX Controller you want to remove has a status of not active.
    nsx-controller-1> get control-cluster status
    uuid: e075cf44-0d49-4eb2-9e4f-d8b10ca97a3b
    is master: true           
    in majority: true           
    uuid                                 address              status              
    06996547-f50c-43c0-95c1-8bb644dea498 192.168.110.53       active              
    471e5ac0-194b-437c-9359-564cea845333 192.168.110.54       active              
    e075cf44-0d49-4eb2-9e4f-d8b10ca97a3b 192.168.110.51       active              
    863f9669-509f-4eba-b0ac-61a9702a242b 192.168.110.52       not active
  4. Detach the controller from the cluster.
    nsx-controller-1> detach control-cluster 192.168.110.52
    Successfully detached node from the control cluster.
    
  5. Detach the controller from the management plane.
    nsx-manager-1> detach controller 863f9669-509f-4eba-b0ac-61a9702a242b
    The detach operation completed successfully
  6. Verify the controllers are active and the control cluster is stable.

    From an NSX Controller:

    nsx-controller-1> get control-cluster status 
    uuid: e075cf44-0d49-4eb2-9e4f-d8b10ca97a3b
    is master: true           
    in majority: true           
    uuid                                 address              status              
    06996547-f50c-43c0-95c1-8bb644dea498 192.168.110.53       active              
    471e5ac0-194b-437c-9359-564cea845333 192.168.110.54       active              
    e075cf44-0d49-4eb2-9e4f-d8b10ca97a3b 192.168.110.51       active

    From an NSX Manager:

    nsx-manager-1> get management-cluster status
    Number of nodes in management cluster: 1
    - 192.168.110.201  (UUID 4213216E-F93A-71B2-DA20-AFE5E714644F) Online     
    - 192.168.110.202  (UUID 4227F3D2-B7FE-8925-EA45-95ECD829C3E2) Online          
    - 192.168.110.203  (UUID 4227824A-1BDD-3A72-3EB3-8D306FEAE42D) Online      
    
    Management cluster status: STABLE
    
    Number of nodes in control cluster: 3
    - 192.168.110.51   (UUID e075cf44-0d49-4eb2-9e4f-d8b10ca97a3b)
    - 192.168.110.53   (UUID 06996547-f50c-43c0-95c1-8bb644dea498)
    - 192.168.110.54   (UUID 471e5ac0-194b-437c-9359-564cea845333)
    
    Control cluster status: STABLE
    

Results

Note:

The controller that was removed with the detach command still retains some configuration information. If you want to join the controller again to any controller cluster, you must run the following CLI command on the controller to remove the stale information:

    deactivate control-cluster