A tornado based sample python webserver in a docker image

web-server.py is a sample python application that hosts a simple webserver that responds to Get requests http requests on localhost:8888 with the hostname.

To run the webserver manually,

1. Install python versions 2.7 or 3.3+.
2. Install Tornado - pip install tornado
3. Run python web-server.py
4. Try accessing http://localhost:8888 from your browser.

Sharing and running such a simple app like this web-server requires many manual steps. Imagine packaging a real-life webserver with a lot of web content.

Now, let's see how docker can help. Take a look at the Dockerfile before you begin. Docker allows for stacking of images on top of each other. Our docker image will be built on top of an existing docker image library/python which has python pre-installed.

1. Install Docker following these instructions.

2. Execute docker build -t "py-web-server:v1" . to build a docker image with web-server packaged inside of it.

3. Run the webserver

docker run -d -p 8888:8888 -h my-web-server py-web-server:v1

The webserver and all its dependencies including python and tornado library have been packaged into a single docker image that can now be shared with everyone. py-web-server:v1 docker image will function the same way on all docker supported OSes (OS X, Windows & Linux).

Try accessing the webserver again - http://localhost:8888

Now let's upload the docker image to your private image repository in Google Cloud Registry (gcr.io).

1. Store the GCP project name in an environment variable

export GCP_PROJECT=<your-project-name>

2. Rebuild the docker image with an image name that includes gcr.io project prefix.

docker build -t "gcr.io/${GCP_PROJECT}/py-web-server:v1" .

3. Push the image to gcr.io

gcloud docker push -- gcr.io/${GCP_PROJECT}/py-web-server:v1

The image is now private and people with access to $GCP_PROJECT used above can only access it.

Now, let's make the image accessible to everyone. Google Container Registry stores its images on Google Cloud storage. Let's update the permissions on Google Cloud Storage to make our image repository publically accessible.

gsutil defacl ch -u AllUsers:R gs://artifacts.${GCP_PROJECT}.appspot.com
gsutil acl ch -r -u AllUsers:R gs://artifacts.${GCP_PROJECT}.appspot.com
gsutil acl ch -u AllUsers:R gs://artifacts.${GCP_PROJECT}.appspot.com

The docker image can now be executed from any machine by running the following command

docker run -d -p 8888:8888 -h my-web-server gcr.io/${GCP_PROJECT}/py-web-server:v1

Detailed Dockerfile reference is available here.