Creating a Docker image based on another image

In this tutorial we will go over how to make changes to a running container while attached, then saving our progress. This will result in a new deployable image.

Install the nginx server to your container and commit the changes.

  1. First attach to the running CentOS container, like we did in the last tutorial.
    docker attach <container name>

    (You can use either container name or container id number. Look at the running container and find this information with docker ps)

  2. Install nginx via yum. You will also need to install the epel-release package to get access to nginx.

    yum install epel-release
    yum update
    yum install nginx
  3. Now that nginx is installed, exit out of the container with the following key combination:
    CTRL` + `P` and `CTRL` + `Q`
  4. Commit your changes to a new image with the following command:
    docker commit -m 'added nginx' --change='CMD ["nginx", "-g daemon off;"]' <container name> dopensource/nginx
  • Notice the similarities in syntax to git.
  • -m 'added nginx' creates a comment for this commit.
  • --change='CMD ... is changing the CMD command, which is what the image will run when it is first started up. In this example, we are telling the image to run nginx in the forground. Most base os images have CMD set to bash so you can interact with the os when attaching.
  • <container name> is the name of the container you want to commit from. You can again, get this from docker ps
  • dopensource/nginx is our name for the new image. dopensource is our organization and nginx is the name of the container.
  1. You can now view your new image with the following command:
    docker images


  • Notice there are two images here. One for the base CoreOS image, which we pulled from the docker hub, and the image we just created.

Run the new image

docker run -dit dopensource/nginx

In the next tutorial we will go over how to allow access to the container over the network.