Adam the Automator

How To Create A Docker Windows Image Using Docker Build

Kevin Sapp

Read more posts by this author.

Are you new to Docker Windows Images? Are you currently working in a Windows shop and curious to learn about Docker builds for container images? You have come to the right place. The best way to learn about new something is by doing.

In this article, you are going to learn how to create your first Windows Docker image from a Dockerfile using the docker build command.

Let’s get started!

Table of Contents

Understanding Docker Container Images

For years, the only way to test or perform development on multiple operating systems (OS) was to have several dedicated physical or virtual machines imaged with the OS version of your choice. This methodology required more hardware and overhead to provision new machines for each software and OS specification.

However, these days the usage of Docker container images has grown partly due to the popularity of micro-service architecture. In response to the rise in Docker’s popularity, Microsoft has started to publicly support Docker images for several flagship products on their Docker Hub page. They have even added native support for images for Windows as a product feature in the Windows 10 and Windows Server 2016!

A Docker image is run on a container by using the Docker Engine. Docker images have many benefits such as portability (applicable to multiple environments and platforms), customizable, and highly scalable. As you can see below, unlike traditional virtual machines, the Docker engine runs on a layer between the host OS kernel and the isolated application services that are being containerized.

Understanding Docker Build and Images

The docker build **command can be leveraged to automate container image creation, adopt a container-as-code DevOps practice, and integrate containerization into the development cycle of your projects. Dockerfiles are simply text files that contain build instructions used by Docker to create a new container image that is based on an existing image.

The user can specify the base image and list of commands to be run when a container image is deployed or startup for the first time. In this article, you will learn how to create a Windows-based docker image from Dockerfile using a Windows container.

This process has several benefits over using a pre-built container image:

1. You are able to rebuild a container image for several versions of Windows – which is great for testing code changes on several platforms.
2. You will have more control over what is installed in the container. This will allow you to keep your container size to a minimum.
3. For security reasons, you might want to check the container for vulnerabilities and apply security hardening to the base image

Prerequisites/Requirements

This article is a walkthrough on learning about learning how to build a Docker image using a Dockerfile. If you’d like to follow along, ensure that you have the following prerequisites in place.

• Docker for Windows installed. I’ll be using the Docker Community Edition (CE) version 2.1.0.4 in my environment.
• Internet access is needed for downloading the Docker images
• Windows 10+ Operating System (version 1709 is being used for this tutorial)
• Nested virtualization enabled
• 5 GB of free diskspace on your local machine
• PowerShell 5.0+
• This tutorial uses the Visual Studio Code IDE. However feel free to use what ever IDE you’d prefer.

Note: Be sure to enable Windows Containers Configuration when installing Docker.

Getting Prepared

You’ll first need a folder to store all of the Docker images and containers you’ll be building from those images. To do so, open a Powershell or cmd terminal (you’ll be using PowerShell throughout this article) and create a new directory called C:\Containers.

Once the folder is created, change to that directory. This puts the console’s current working directory to C:\Containers to default all downloads to this directory.

In this article, you’ll get a headstart. Most of the files to work through this project are already available. Once the folder is created, perform a Git pull to copy over the files needed for this article from the TechSnips Github repository to the C:\Containers folder. Once complete, check to make sure that the C:\Containers folder looks like below.

Downloading the IIS Windows Docker Image

The first task to perform is to download a “template” or base image. You’ll be building your own Docker image later but first, you need an image to get started with. You’ll be downloading the latest IIS and Windows Server Core Images that are required for this tutorial. The updated list of images can be found on the official Microsoft Docker hub image page.

Reviewing the Current Docker Base Images

Before downloading the image from the image repository, let’s first review the current Docker base images that you currently have on your local system. To do so, run a PowerShell console as Administrator and then type docker images . This command returns all images on your local system.

As you can see below, the images available are initially empty.

Downloading the Base Image

Now it’s time to download the base IIS image from Docker Hub. To do so, run docker pull as shown below. This process can take some time to complete depending on your internet speeds.

Now run docker images and you should have the latest Microsoft Windows Core IIS image available for this tutorial.

Inspecting the Dockerfile

In an earlier step, you had downloaded an existing Dockerfile for this tutorial. Let’s now take a look at exactly what that entails.

Open the C:\Containers\Container1\Dockerfile file in your favorite editor. The contents of this Dockerfile are used to define how the container image will be configured at build time.

You can see an explanation of what each piece of this file does in the in-line comments.

Building a New Docker Image

You’ve got the Dockerfile ready to go and a base IIS image downloaded. Now it’s time to build your new Docker image using the Dockerfile.

To build a new image, use the docker build command. This command creates the image. For this article, you can see below you’re also using the -t **option. This option allows you to give your new image a friendly tag name and also reference the Dockerfile by specifying the folder path where it resides.

Below you can see an example of ensuring the console is in the C:\Containers directory and then building a new image from the Dockerfile in the C:\Containers\Container1 directory.

Once started, you can see the progress of the command as it traverses each instruction in the docker file line by line:

Once done, you should now have a new Docker image!

Now run the docker images command to view the images that are available. You can see below an example of the container1 image created.

Note: The docker build —help command is a useful parameter to display detailed information on the docker command being run.

Running the Docker Container

At this point, you should have a new image created. It’s time to spin up a container using that image. To bring up a new container, use the docker run command.

The docker run command will bring up a new Docker container based on the container1 image that you created earlier. You can see an example of this below.

Notice that the -d parameter is used. This tells the docker runtime to start the image in the detached mode and then exit when the root process used to run the container exits.

When docker run completes, it returns the ID of the container created. The example below is capturing this ID into a $containerID variable so we can easily reference it later.

Once the container is brought up, now run the docker ps command. This command allows you to see which containers are currently running using each image. Notice below that the running image is automatically generated a nickname (busy_habit in this case). This nickname is sometimes used instead of the container ID to manage the container.

Running Code Inside a Docker Container

A new container is built from a new image you just created. Let’s now start actually using that container to run code. Running code inside of a Docker container is done using the docker exec command.

In this example, run docker exec to view PowerShell output for the Get-ChildItem command in the container using the command syntax below. This will ensure the instructions in the Dockerfile to remove the default IIS files succeeded.

You can see below that the only file that exists is index.html which means the default files were removed.

Now run the ipconfig command in the container to get the local IP address of the container image so that you can try to connect to the IIS website.

You can see below that ipconfig was run in the container just as if running on your local computer and has return all of the IP information.

Inspecting the IIS Website

Now it’s time to reveal the fruits of your labor! It’s time to see if the IIS server running in the Docker container is properly serving up the index.html page.

Open a browser and paste the IP4 Address found via ipconfig into the address bar. If all is well, you should see a Hello World!! message like below.

Reviewing Docker History

One useful command to use when working with Docker containers i the docker history command. Although not necessarily related to creating an image or container itself, the docker history command is a useful command that allows you to review changes made to the container image.

You can see below, that docker history returns all of the Dockerfile and PowerShell activity performed on the container1 container you’ve been working with.

Cleaning up the Running Docker Images

The steps below are used to cleanup all stopped containers running on your machine. This will free up diskspace and system resources.

Run the docker ps command to view a list of the containers running on your system:

Now stop the running containers using the docker stop command:

Finally you can permanently remove the stopped containers using the docker system prune command.

Docker hub windows image

This repository contains build instructions and Dockerfile to build Docker images with Windows-only browsers: Internet Explorer and Microsoft Edge .

1. Bare metal machine or on VM with nested virtualization enabled and Linux installed. This example was tested on Ubuntu 18.04 .

To check that virtualization is supported — verify that /dev/kvm file is present:

1. Qemu machine emulator installed. It is important to use the same qemu version on host machine where images are built and inside Docker image. To check qemu version type:

1. Windows license key

1. Preparative Steps

1.1) Clone this repository and change dir to it:

1.2) Download Windows 10 installation image from Microsoft Software Download website. 1.3) Download virtio drivers virtio-win-0.1.141.iso. In the next steps we assume that you now have two files in current directory:

2. Windows Installation

2.1) Create hard disk image where Windows will be installed:

2.2) Run virtual machine and begin installation:

2.3) Windows will boot from installation image. Install Windows.

2.3.1) Proceed to the next step:

2.3.2) Click Install now:

2.3.3) Enter license key:

2.3.4) Choose Windows edition:

2.3.5) Read and accept license agreement:

2.3.6) Choose custom installation type:

2.3.7) Now you have to install virtio storage driver. Click Load driver:

2.3.8) Point to E:\viostor\w10\x86 directory:

2.3.9) Click next to install driver:

2.3.10) Choose installation partition and click next:

2.3.11) Wait while installation finishes:

2.3.12) Setup user and password:

2.3.13) Do other post-install configuration steps until you get Windows installed:

2.3.14) Install Ethernet virtio driver. Open device manager and click Update driver: Choose virtio cdrom and click OK: Install driver: Connect to network:

2.3.15) Disable Windows Firewall or add firewall rule to allow access to port 4444. This is needed to access webdriver binary port with Selenium test.

2.3.16) Configure Windows as you wish: install updates, change screen resolution, apply registry modifications and so on.

3. Adding WebDriver Binaries

These binaries will handle Selenium test requests and launch respective browser.

For Internet Explorer — download an archive with driver binary from Selenium official website, unpack it and put the binary to C:\Windows\System32 directory.

For Microsoft Edge web driver binary can be installed with the following command:

4. Creating Quick Boot Memory Snapshot

This snapshot contains memory state and is needed to quickly restore virtual machine instead of doing full boot which is slow. To create it:

4.1) Shutdown virtual machine.

4.2) Create overlay image that will contain VM state:

4.3) Run VM using snapshot.img as filesystem:

Please note that qemu runs with monitor connected to stdio.

4.4) Browser configuration (required only for Internet Explorer).

Open Internet Explorer. The first time this browser is launched, it asks for the security setup. The option «Don’t use recommended settings» need to be selected as follows:

Then, the Internet Options have to be changed. These options can be opened using the configuration button located at the top of Internet Explorer. In the tab «Security», the protect mode for the zones «Internet» and «Restricted sites» have to be disabled, as shown in the following picture:

At this point, you have to close Internet Explorer. Select the option «Always close all tabs» when Internet Explorer is closing. Finally, you have to open again Internet Explorer and double check that the protected mode is turned off (it can be seen in a message box at the bottom of the browser).

4.5) Run web driver binary command.

• For Microsoft Edge — open command prompt with administrator privileges and run:

• For Internet Explorer — open command prompt as unprivileged user and run:

4.6) Minimize command line prompt window when driver is up and running.

4.7) Switch to terminal where qemu runs and type at qemu prompt:

Then type quit to stop VM:

To start VM from snapshot manually use the following command:

The command above is used in Dockerfile entry point script.

5. Build Docker Image

5.1) Move filesystem and state files to image directory in this repository:

5.2) Build Docker image using provided Dockerfile:

For Internet Explorer use:

5.3) Run a container from image:

5.4) To see Windows screen inside running container — connect to vnc://localhost:5900 using selenoid as password.

5.5) To run Selenium tests — use http://localhost:4444 as Selenium URL.

About

Step by step instructions for building Docker images with Windows