Step 2: Configuring the Load Balancer

Important:

This step is mandatory for a multi-node HA-ready production deployment or a single-node evaluation deployment with a dedicated Task Mining and/or GPU node.

If you are configuring a single-node evaluation deployment without a dedicated Task Mining and/or GPU node, proceed to Configuring Azure SQL.

If you are using the Azure Internal Load Balancer (LB) for deployments, you can encounter issues with the calls from the backend Virtual Machine (VM) to the LB frontend IP. The issues occur due to source IP and MAC address mismatch of the network packet. This prevents the recipient from working out the correct response path, resulting in the failure of calls from the VM to the LB. For more details, see Azure Load Balancer Components limitations and Backend Traffic Troubleshooting.

Step 2.1: Adding a frontend IP

1. Create an Azure load balancer using this link. Click Create.
   
   

   

   
   
2. Add to the same Resource Group and give the instance a name.
   
   

   

   
   
3. Go to Next: Frontend IP configuration.
4. Click Add a frontend IP.
   
   

   

   
   
5. Click Add.

Step 2.2: Creating the backend pool and add nodes

Note:

If you are installing Task Mining, the dedicated Task Mining should not be added in the node pool.

The Task Mining Analyzer node should not be added in the node pool.

1. Create the Backend Pool.
   
   

   

   
   
2. Click on Add under Virtual Machines.
   
   

   

   
   
3. You can now proceed with the creation of the Load Balancer by clicking Review + create.
   
   

   

   
   

Make sure to repeat these steps for all virtual networks corresponding to each node.

The recommended configuration is to add two backend pools, as follows:

• one backend pool that includes all server nodes (referred to as the Server Pool);
• one backend pool with all server and agent nodes (minus the Task Mining node, and this pool is referred to as the Node Pool in our documentation).

The Server Pool is associated with the kubeapi-probe, whereas the Node Pool is associated with the https-probe.

Step 2.3: Adding health probes

1. Go the Health Probes section in the Settings tab, and click Add.
   
   

   

   
   
2. Add these 3 health probes for the load balancer.

   Go to health probe and click Add, filling in the the configurations below:

   • https-probe:

     • protocol TCP

     • port 443

   • kubeapi-probe:

     • protocol TCP

     • port 6443
3. For each health probe, fill in the following configuration form:
   

Step 2.4: Adding a load balancing rule

The load balancer must forward traffic to all nodes on the ports 443, 6443, and 9345.

1. Go to the Load balancing rules in the Settings tab.
   
   

   

   
   
2. Create three load balancing rules for each of the health probes:

   • https-probe

     • 443 - HTTPS

   • kubeapi-probe

     • 6443 - Kube API

   See the following configuration page for updating the load balancing rule:

   
   

   

   
   

   For specifying the backend pool, make sure you add the kubeapi-probe to the server pool and the https-probe to the node pool.

Step 2.5: Creating a load balancer domain name

1. Search the public IP address assigned to load balancer in the search bar.
2. Fill in the DNS name label and click Save.