Kloudfuse Install Without External Load Balancer (AWS)

Once the Kloudfuse installation is complete, run through the following steps if you don’t want the Kloudfuse stack to run with an external load balancer.

Note: This has been validated for a single-node deployment so far. If you have more than one node in your deployment, please let us know.

  • Get the internal IP of the node by running kubectl get nodes -o wide. Note down the INTERNAL-IP from the output.

  • Edit the kfuse-ingress-nginx-controller-internal service by running kubectl edit service kfuse-ingress-nginx-controller-internal. Change the spec.type to ClusterIP from LoadBalancer. The spec should look like this:

spec: type: ClusterIP
  • Save and quit the edit session.

  • Re-edit the service spec by running kubectl edit service kfuse-ingress-nginx-controller-internal, and add the externalIPs section in spec. You’ll have to add the internal IP of the node (see first bullet point) here. The spec should now look like:

spec: type: ClusterIP externalIPs: - <Node_Internal_IP> clusterIPs: - <Cluster_IP>

Do not change/modify the clusterIPs in the spec, just add the externalIPs section.

  • Update the security groups on your EC2 instance (from the AWS console) to allow https traffic for the IP, which was just added. See the screenshots below on how to do it from the EC2 instance landing page.

You should be able to access Kloudfuse UI now (using the node’s external IP). And this will not go through the load balancer.

Do not use the node’s internal IP to access the Kloudfuse UI. You can only access the Kloudfuse UI using the external IP.