Expose your Kubernetes resources to the world!
You have created a Kubernetes cluster at home with k3sup, or you are using a Minikube instance. You have created your Pod
, your Deployment
or Service
and your are ready to publicly test it.
Exposing your deployment to the Internet is not so trivial though!
To install net-expose
through Helm you need to add this git repository as a helm repo:
helm repo add net-expose https://raw.githubusercontent.com/operatorequals/k8s-net-expose/master/charts/stable
helm repo update
helm install nginx-ingress net-expose/net-expose \
--set ssh.host=ssh-connect.example.com \
--set ssh.user=admin \
--set-file ssh.key=~/.ssh/id_rsa \
--set remotePort=8080 \
--set k8sExportService.name=nginx-ingress-controller \
--set k8sExportService.port=80
Exposes internal Kubernetes resources (anything with an IP and port) to public VPS hosts.
You only need a working SSH key for the VPS. The rest is on net-expose
.
The net-expose
deployment consists of a Kubernetes Pod
with just enough privileges to remotely SSH to a VPS and establish a
reverse SSH proxy, redirecting VPS port traffic to the internal Kubernetes resource (Pod
or Service
) to be exposed,
making it accessible through the initial VPS port.
OpenSSH server does not allow reverse tunnels to expose ports to all interfaces by default (only to localhost
). To enable this
the GatewayPorts
has to be set to yes
in the /etc/ssh/sshd_config
as below:
[...]
GatewayPorts yes
[...]
To automate this process the gatewayports_test.sh
can be used as below:
ssh -i id_rsa REMOTE_HOST -l root ' sudo sh -s ' < gatewayports_test.sh
To apply the change the SSH server needs a restart.
A K8s initContainer
can be created from the Helm release to automatically run the gatewayports_test.sh
towards the remote
host and restart the SSH server before creating the connection tunnel.
To enable this feature the gatewayports.enable
needs to be set to true
in the Helm values.yaml
as below:
helm install nginx-ingress net-expose/net-expose \
[...]
--set gatewayports.enable=true