This article explains how to configure a cluster in NSX Advanced Load Balancer for an OpenStack cloud. To provide NSX Advanced Load Balancer Controller high availability (HA), add two additional Controller nodes to create a 3-node Controller cluster. For more details on deploying a cluster, refer to Deploying an NSX Advanced Load Balancer Controller Cluster.

Deploying an NSX Advanced Load Balancer Controller Cluster

For complete information on configuring Controller’s management interfaces and cluster IP, refer to Deploying an NSX Advanced Load Balancer Controller Cluster. The following steps are for creating OpenStack floating IP and binding that with the cluster IP:

Write Mode

Prerequisites

There are certain prerequisites defined for the leader and follower nodes in a cluster. For complete information, refer to Prerequisites for Cluster Deployment.From an OpenStack perspective, consider the following:

  1. A Neutron port is created and is available for cluster VIP.

  2. A floating IP is available for Neutron port.

Procedure

  1. Access OpenStack Horizon CLI
    1. List the Network : openstack network list — This indicates the configured requisite networks.
    2. Create a floating IP : openstack floating ip create provider1 — *provider1* is the network used.
    3. New floating IP is created.
    4. Get the port-id for cluster IP.
    5. Associate the cluster IP with the floating IP: Using the port-id from the command above (95665123-64a4-453a-abde-70fdb3d2ae2a in this case), associate it with the floating IP created in step b.
  2. Add the cluster IP and the secondary IP for the cluster leader. As shown in Figure 1, 172.16.0.65 is the cluster IP which is added as the secondary IP for the cluster leader.
  3. No-Access Mode
    1. For OpenStack No-Access cloud type, the AAP entries need to be configured manually using the following command. An example is shown in the code block below.
    2. Create the neutron port for the VIP by using the following command.
      Note:

      When the leader Controller fails (or reboots), a follower Controller will take over the cluster IP (in this case 172.16.0.65), and the mapping between floating IP (10.130.170.86) and cluster IP (172.16.0.65) will not change. Therefore, without intervention, the floating IP and cluster IP association will work as expected.