There are two ways to change the memory and CPU resources of an NSX Manager node in a cluster.

Note that in normal operating conditions all three manager nodes must have the same CPU and memory resources. A mismatch of CPU or memory between NSX Managers in an NSX Manager cluster should only be done when transitioning from one size of NSX Manager to another size.

If you have configured resource allocation reservation for the NSX Manager VMs in vCenter Server, you might need to adjust the reservation. For more information, see the vSphere documentation.

Option A (resize a manager node with the same IP address) requires less effort. NSX-T requires that two managers are available at all times. If you have cluster VIP (virtual IP) configured, there will be a brief outage when the VIP switches to another node in the cluster. You can access the other two nodes directly during the outage if the VIP-assigned node is shut down for resizing. If you have deployed a load balancer for the manager nodes, health checks will be triggered when a manager goes offline. The load balancer should direct traffic to another node. Choose this option if you do not want to change the IP address of the manager nodes.

For option B (resize a manager node with a different IP address), you will need IP addresses for the three new managers. If you have cluster VIP configured, there will be a brief outage when the VIP switches to another node in the cluster. You can access the other two nodes directly during the outage in case the VIP-assigned node is deleted. If you have deployed a load balancer for the manager nodes, health checks will be triggered when a manager goes offline. The load balancer should direct traffic to another node. After all the steps are completed, you will need to reconfigure the load balancer (add the new managers and remove the old managers).

When you deploy a new manager node from the NSX Manager UI, if you get the error message "The repository IP address ... is not a part of the current management cluster. Please update the repository IP to the current node by running repository-ip CLI command. (Error code: 21029)," log in to the CLI of one of the existing nodes as admin and run the command set repository-ip. This will resolve the error.

Prerequisites

Procedure

  • Option A: Resize a manage node with the same IP address
    Option A.1: Change the CPU and/or memory of the existing manager nodes. You must make the change to one manager at a time so that two managers are available at all times.
    1. Log in to a manager's CLI as admin and run the shutdown command.
    2. From NSX Manager UI, verify that the state of the manager cluster is DEGRADED.
    3. From vSphere, change the memory and/or CPU resources of the manager VM that was shut down.
    4. From vSphere, power on the VM. From NSX Manager UI, wait for the state of the manager cluster to be STABLE.
    5. Repeat steps 1 to 4 for the other two manager VMs.
    Option A.2: Deploy new manager nodes.
    1. From NSX Manager UI, delete a manager node that was deployed from NSX Manager UI.
    2. From NSX Manager UI, deploy a new manager node with the new size into the cluster with an IP address that is the same as the one used by the manager node that was deleted in step 1.
    3. From NSX Manager UI, wait for the state of the manager cluster to be STABLE.
    4. Repeat steps 1 to 3 for the other manager node that was deployed from NSX Manager UI.
    5. For the manually-deployed manager node, log in to its CLI as admin and run the shutdown command.
    6. From another manager node, log in to its CLI as admin and run the get cluster config command to get the node ID of the manually-deployed manager node. Then run the detach node <node-id> command to detach the manually-deployed manager node from the cluster.
    7. From vSphere, delete the manually-deployed manager node VM.
    8. From NSX Manager UI, deploy a new manager node with the new size into the cluster with an IP address that is the same as the one used by the manually-deployed manager node.
    9. From NSX Manager UI, wait for the state of the manager cluster to be STABLE.
  • Option B: Resize a manage node with a different IP address
    1. If you have VIP configured and if the new addresses and old addresses are in different subnets, from NSX Manager UI, remove the VIP.
      You must access NSX Manager using the manager's IP address and not the VIP address.
    2. From NSX Manager UI, deploy a new manager node with the new size into the cluster with an IP address that is different from the ones used by the current manager nodes.
    3. From NSX Manager UI, verify that the state of the manager cluster is STABLE.
    4. From NSX Manager UI, delete an old manager node that was deployed from NSX Manager UI.
    5. Repeat steps 1 to 3 for the other manager node that was deployed from NSX Manager UI.
    6. For the manually-deployed manager node, log in to its CLI as admin and run the shutdown command.
    7. From another manager node, log in to its CLI as admin and run the get cluster config command to get the node ID of the manually-deployed manager node. Then run the detach node <node-id> command to detach the manually-deployed manager node from the cluster.
    8. From vSphere, delete the manually-deployed manager node VM.
    9. From NSX Manager UI, deploy a new manager node with the new size into the cluster with an IP address that is different from the one used by the manually-deployed manager node.
    10. From NSX Manager UI, wait for the state of the manager cluster to be STABLE.
    11. If you removed the old VIP in step 1, from NSX Manager UI, configure a new VIP. It must be in the same subnet that the new IP addresses are in.