Docker Network

Containers run in isolated environments. This means a container is not automatically connected to the same user-defined network as another container, and it does not automatically know which application containers it should communicate with.

Docker does provide a default bridge network. Containers on that network can communicate using IP addresses, but this is less practical because IP addresses can change. To connect several containers in a clearer and easier-to-manage way, create your own Docker Network.

A Docker Network lets containers communicate using container names. This is important when an app needs to connect to a database, cache, or another service.

How containers communicate

Containers on the same Docker user-defined network can connect to each other by name. You do not need to know their IP addresses.

Loading diagram...

localhost inside a container points to the container itself. So, from the app container, access the database in the db container using the hostname db, not localhost.

View Docker Networks

docker network ls

Example output:

NETWORK ID     NAME      DRIVER    SCOPE
7f3a1b2c4d5e   bridge    bridge    local

This output shows Docker's default bridge network. For a project, it is better to create your own network so its name and members are clear.

Create a network

Command format:

docker network create <network-name>

Example:

docker network create app-network

View network details:

docker network inspect app-network

Run containers on the same network

Command format:

docker run -d --name <container-name> --network <network-name> <image:tag>

To demonstrate communication between containers, run PostgreSQL as the first service on app-network:

docker run -d --name db \
  --network app-network \
  -e POSTGRES_PASSWORD=secret \
  postgres:16

INFO

The -e option adds an environment variable to the container. POSTGRES_PASSWORD=secret sets PostgreSQL's password. See lesson 8 for a fuller explanation of environment variables.

Now run a temporary container on the same network. We use BusyBox, a small Linux image that contains common basic utilities. It fits this example because it provides nslookup, a command that checks whether the name db can be found on the network.

--rm tells Docker to remove the BusyBox container after the command finishes. This container is only a tool for testing the connection, not an application service.

docker run --rm \
  --network app-network \
  busybox nslookup db

The second container can find db by its container name. In an app that connects to PostgreSQL, use host db and port 5432.

Network and ports

Docker Network controls communication between containers. Port mapping with -p controls access from the host, such as access from a laptop browser to a container.

You do not need to publish a database port to the laptop if only an app on the same network uses it. Publish a port only when the host needs access.

After confirming that communication works, remove the practice container and network. These commands do not remove the PostgreSQL image:

docker rm -f db
docker network rm app-network

Docker Compose in the next lesson will create a project network automatically. The concept remains the same: services can reach each other by service name.