When dealing with your own Kubernetes Cluster (not hosted in GCE or AWS) you will have trouble to exposes automatically public services. With these cloud providers, we simply have to create a LoadBalancer service and just wait a bit to have an address automatically added in the status of the service.

This load-balancer profixier brig this awesome workflow to self-hosted Kubernetes cluster, by automatically create kubernetes-reverse-proxy configurations for LoadBalancer services that do not have this configuration. Once the configuration is created, it updates the service status to add the expected DNS address.

First of all, if not already did setup kubernetes-reverse-proxy on one or more of your cluster nodes. The easiest way is to simply run the following command:

docker run -d -t \
    -e CONFD_ETCD_NODE=[YOUR-ETCD-IP]:2379 \
    -p 80:80 \
    --restart=always \
    --name=reverse-proxy \
    darkgaro/kubernetes-reverseproxy

Then, start the profixier on one of your nodes (the master might be the more relevant one):

docker run -d \
    --restart=always \
    --name=load-balancer-proxifier \
    -e ROOT_DNS_DOMAIN=any.wildcarded.dns.address \
    -e CLUSTER_ADDRESS=https://username:password@your.master.cluster.address \
    -e INSECURE_CLUSTER=true \
    sroze/kubernetes-load-balancer-proxifier

The proxifier is listening on all events related to services. Once one is created or updated, if no kubernetesReverseproxy annotation is found, it'll automatically create one based on the root DNS domain you choose and send this created DNS address back to the service status.