Shutdown/Startup order of the NSX for vSphere 6.x environment after a maintenance window or a power outage

Products

• VMware NSX Networking

Issue/Introduction

This article provides the order in which VMware NSX for vSphere 6.x objects should be powered off and on during a maintenance window or during a power outage.

Note: This article assumes that your physical network and SAN fabric are in working order and are not experiencing any other issues.

Environment

VMware NSX for vSphere 6.2.x
VMware NSX for vSphere 6.0.x
VMware NSX for vSphere 6.1.x
VMware NSX for vSphere 6.4.x
VMware NSX for vSphere 6.3.x

Resolution

Power Off Sequence

Note: VMware recommends to back up the NSX environment before initiating any planned power-off exercises. For more information, see the Backup and Restore section in the NSX Administration Guide.

1. To protect the integrity of the NSX environment, NSX Manager should be shut down to prevent the administrators or an application from making a change to the environment when one or more NSX objects are already deleted.
   In case of Multi vCenter environment, follow this sequence:
   • Power off Secondary NSX Manager.
   • Power off Primary NSX Manager.
   • Power off Primary NSX Controller.
2. To prevent new workloads from initiating new data plane connections, shut down the DLR control VM next.
    
3. After completing the first two steps, the rest of the NSX objects, including all virtual machines, Guest Introspection VM, and any installed third-party services and their component virtual machines can be shut down in any order through the vCenter Server User Interface (UI).

Note: When performing a complete shutdown, all controller VMs can be shutdown at the same time, instead of one at a time. Do not use the Power Off option. Instead, use either Shutdown Guest (if available through VMware Tools) or the documented shutdown process for the specific Guest OS.

Power On Sequence

The power-on sequence is more prescriptive.

1. Power on vCenter Server and ensure that the initialization completes.
2. Power on NSX Manager and ensure all NSX services are up. For example, ensure that NSX Manager has connected to the lookup service.
   (In case of Multi vCenter environment, Power on Primary NSX Manager first and then Secondary NSX Manager followed by Primary NSX Controller).
3. Power on ESXi hosts and ensure that the initialization completes.
   • Ensure that Guest Introspection and NSX Edges boot up automatically.
     Note: Ensure to complete this step before powering on the controllers to avoid the controller power-on process hanging indefinitely while it waits for the service VMs to initialize.
4. Power on the NSX controllers and ensure that the controllers forms a cluster majority.
   Note: You must power-on all of the controllers at the same time. Do not wait for the primary controller to fully initialize before powering on the other controllers. The controller status does not change to Normal until all other controllers are powered On. It is expected that the status of all controllers should display the status of Normal.
5. To monitor controller status, log in through the vSphere Web Client and navigate to Networking and Security > Installation > Management. The status of the first controller changes to Normal after the secondary controller is powered on and changes to Normal as well.
6. Ensure that the ESXi hosts can connect to the controllers.
7. Power on the Distributed Logical Router and ensure that the forwarding plane is working as expected.
8. Power on the virtual machines and test network connectivity.
   Note: This procedure delays the virtual machine start-up after all NSX components have started to avoid application failures because network connectivity is not established.

Additional Information