Skip to content

Latest commit

 

History

History
136 lines (89 loc) · 8.52 KB

docker_networking.md

File metadata and controls

136 lines (89 loc) · 8.52 KB
Raw

Docker Networking

Docker network sandbox and drivers

It's time to face hard questions about container networking. The core idea of containers is isolation, so how can a container communicate with other container over the network? How can a container communicate with the public internet? It must be using the host machine network resources. How can that be implemented while keeping the host machine secure enough?

Docker implement a virtualized layer for container networking, enables communication and connectivity between Docker containers, as well as between containers and the external network. It includes all the traditional stack we know - unique IP address for each container, virtual network interface that the container sees, default gateway and route table.

Below is the virtualized model scheme. In Docker, this model is implemented by libnetworking library:

A sandbox is an isolated network stack. It includes Ethernet interfaces, ports, routing tables, and DNS configurations.

Network Interfaces are virtual network interfaces (E.g. veth ). Like normal network interfaces, they're responsible for making connections between the container and the rest of the world.

Network interfaces are connected the sandbox to networks.

A Network is a group of network interfaces that are able to communicate with each-other directly. An implementation of a Network could be a Linux bridge, a VLAN, etc.

This networking architecture is not exclusive to Docker. Docker is based on an open-source pluggable architecture called the Container Network Model (CNM).

The networks that containers are connecting to are pluggable, using network drivers. This means that a given container can communicate to different kind of networks, depending on the driver. Here are a few common network drivers docker supports:

  • bridge: This network driver connects containers running on the same host machine. If you don't specify a driver, this is the default network driver.

  • host: This network driver connects the containers to the host machine network - there is no isolation between the container and the host machine, and use the host's networking directly.

  • overlay: Overlay networks connect multiple container on different machines, as if they are running on the same machine and can talk locally.

  • none: This driver disables the networking functionality in a container.

The Bridge network driver

The Bridge network driver is the default network driver used by Docker. It creates an internal network bridge on the host machine and assigns a unique IP address to each container connected to that bridge. Containers connected to the Bridge network driver can communicate with each other using these assigned IP addresses. The driver also enables containers to communicate with the external network through port mapping or exposing specific ports.

The default bridge network official tutorial demonstrates how to use the default bridge network that Docker sets up for you automatically.

The user-defined bridge networks official tutorial shows how to create and use your own custom bridge networks, to connect containers running on the same host machine.

Complete both Use the default bridge network and Use user-defined bridge networks tutorials.

The Host network driver

The Host network driver is a network mode in Docker where a container shares the network stack of the host machine. When a container is run with the Host network driver, it bypasses Docker's virtual networking infrastructure and directly uses the network interfaces of the host. This allows the container to have unrestricted access to the host's network interfaces, including all network ports. However, it also means that the container's network stack is not isolated from the host, which can introduce security risks.

Complete Docker's short tutorial that demonstrates the use of the host network driver:
https://docs.docker.com/network/network-tutorial-host/

IP address and hostname

By default, the container is allocated with an IP address for every Docker network it attaches to. A container receives an IP address out of the IP pool of the network it attaches to. The Docker daemon effectively acts as a DHCP server for each container. Each network also has a default subnet mask and gateway.

As you've seen in the tutorials, when a container starts, it can only attach to a single network, using the --network flag. You can connect a running container to multiple networks using the docker network connect command.

In the same way, a container's hostname defaults to be the container's ID in Docker. You can override the hostname using --hostname.

DNS services

Containers that are connected to the default bridge network inherit the DNS settings of the host, as defined in the /etc/resolv.conf configuration file in the host machine (they receive a copy of this file). Containers that attach to a custom network use Docker's embedded DNS server. The embedded DNS server forwards external DNS lookups to the DNS servers configured on the host machine.

Custom hosts, defined in /etc/hosts on the host machine, aren't inherited by containers. To pass additional hosts into container, refer to add entries to container hosts file in the docker run reference documentation.

The EXPOSE Dockerfile instructions

The EXPOSE instruction informs Docker that the container listens on the specified network ports at runtime.

The EXPOSE instruction does not actually publish the port. It functions as a type of documentation between the person who builds the image and the person who runs the container, about which ports are intended to be published. To actually publish the port when running the container, use the -p flag on docker run to publish and map one or more ports.

Self-check questions

Enter the interactive self-check page

Exercises

✏️ Inspecting container networking

Run the busybox image by:

docker run -it busybox /bin/sh
  1. On which network this container is running?
  2. What is the name of the network interface that the container is connected to, as it is seen from the host machine?
  3. What is the name of the network interface that the container is connected to as it is seen from within the container?
  4. What is the IP address of the container?
  5. Using the route command, what is the default gateway ip that the container use to access the internet?
  6. Provide an evidence that the container's default gateway IP is the IP address of the default bridge network on the host machine the container is running on.
  7. What are the IP address(es) of the DNS server the container used to resolve hostnames? Provide an evidence that they are identical to the DNS servers host machine.
  8. Create a new bridge network, connect your running container to this network.
  9. Provide an evidence that the container has been connected successfully to the created network.
  10. From the host machine, try to ping the container using both its IP addresses.
  11. After you've connected the container to a custom bridge network, what are the IP address of the DNS server the container used to resolve hostnames? What does it mean?

✏️ Flask, Nginx, MongoDB

Your goal is to build the following architecture:

  • The Dockerfiles of the nginx and the flask apps can be found in our shared repo under nginx_flask_mongodb.
  • The mongo app should be run using the pre-built official Mongo image.
  • The nginx and flask app should be connected to a custom bridge network called public-net-1 network.
  • In addition, the flask app the mongo should be connected to a custom bridge network called private-net-1 network.
  • The nginx should talk with flask using the flask-app hostname.
  • The flask app should talk to the mongo using the mongo hostname.