Skip to content

Dual tunnels

Setting up dual TCP and HTTPS tunnels

In this tutorial we will set both a dual tunnel for exposing HTTP and TCP services from the same server.

Whilst it's easier to automate two separate servers or cloud instances for your tunnels, you may want to reduce your costs.

The use-case may be that you have a number of OpenFaaS functions running on your Raspberry Pi which serve traffic to users, but you also want to connect via SSH and VNC.


  • A Linux server, Windows and MacOS are also supported
  • The inlets-pro binary at /usr/local/bin/
  • Access to a DNS control plane for a domain you control

Create the HTTPS tunnel server first

Create a HTTPS tunnel server using the manual tutorial or automated tutorial.

Once it's running, check you can connect to it, and then log in with SSH.

You'll find a systemd service named inlets-pro running the HTTPS tunnel with a specific authentication token and set of parameters.

Now, generate a new systemd unit file for the TCP tunnel.

I would suggest generating a new token for this tunnel.

TOKEN="$(head -c 32 /dev/urandom | base64 | cut -d "-" -f1)"

# Find the instance's public IPv4 address:
PUBLIC_IP="$(curl -s"

Let's imagine the public IP resolved to which is part of the DigitalOcean range.

inlets-pro tcp server \
 --token "$TOKEN" \
 --auto-tls-san $PUBLIC_IP \
 --generate=systemd > inlets-pro-tcp.service


Description=inlets PRO TCP Server

ExecStart=/usr/local/bin/inlets-pro tcp server --auto-tls --auto-tls-san= --control-addr= --token="k1wCR+2j41TXqqq/UTLJzcuzhmSJbU5NY32VqnNOnog=" --control-port=8124 --auto-tls-path=/tmp/inlets-pro-tcp


We need to update the control-port for this inlets tunnel server via the --control-port flag. Use port 8124 since 8123 is already in use by the HTTP tunnel. Add --control-port 8124 to the ExecStart line.

We need to add a new flag so that generated TLS certificates are placed in a unique directory, and don't clash. Add --auto-tls-path /tmp/inlets-pro-tcp/ to the same line.

Next install the unit file with:

sudo cp inlets-pro-tcp.service /etc/systemd/system/
sudo systemctl daemon-reload
sudo systemctl enable inlets-pro-tcp.service

sudo systemctl restart inlets-pro-tcp.service

You'll now be able to check the logs for the server:

sudo journalctl -u inlets-pro-tcp

Finally you can connect your TCP client:

inlets-pro tcp client \
  --token "k1wCR+2j41TXqqq/UTLJzcuzhmSJbU5NY32VqnNOnog=" \
  --upstream \
  --ports 2222,5900 \
  --url wss://

Note that 5900 is the default port for VNC. Port 2222 was used for SSH as not to conflict with the version running on the tunnel server.

You can now connect to the public IP of your server via SSH and VNC:

For example:

ssh -p 2222 pi@

Wrapping up

You know have a TCP and HTTPS tunnel running on the same server by changing the control-plane port and auto-TLS path of the second server and installing a new systemd service.

This technique may save you a few dollars per month, but it may not be worth your time compared to creating two separate servers with inletsctl create. Only you can decide.