Be specific when binding ports!

Most Docker tutorials I’ve seen, mention to bind ports by simply specifying -p 80. This will work for most people getting started with containers. However, I when I was trying to run a Pi-hole docker container, it didn’t start… I restarted my computer and tried again, unsure what other program would’ve been binding to port 53. And this time the container started!… But Waydroid didn’t with the following error dnsmasq: failed to create listening socket for 192.168.250.1: Address already in use

I was confused. How was I ever supposed to run my Pi-hole container at the same time as I get to run android apps through Waydroid?! Since “port 53” would be occupied, right?

Several services on the “same port”?

This is where we have to analyze the error message more closely. Why does it mention an ipaddress and not just the port number? It’s because when binding a port, you bind to an interface. Our Docker port bindings are too broad, we can be even more specific about what we want to bind! -p 192.168.0.2:53:53/udp is the actual specific binding I should pass to Docker. By omitting the interface and protocol you’ll be binding 0.0.0.0:53:53/tcp and :::53:53/tcp (IPv6) by default.