Deployment with Docker


  • Docker 17.05+.
  • Docker Compose 1.17+

Understanding the Docker Compose Setup

Before you begin, check out the production.yml file in the root of this project. Keep note of how it provides configuration for the following services:

  • django: your application running behind Gunicorn;
  • postgres: PostgreSQL database with the application’s relational data;
  • redis: Redis instance for caching;
  • traefik: Traefik reverse proxy with HTTPS on by default.

Provided you have opted for Celery (via setting use_celery to y) there are three more services:

  • celeryworker running a Celery worker process;
  • celerybeat running a Celery beat process;
  • flower running Flower.

The flower service is served by Traefik over HTTPS, through the port 5555. For more information about Flower and its login credentials, check out Celery Flower instructions for local environment.

Configuring the Stack

The majority of services above are configured through the use of environment variables. Just check out Configuring the Environment and you will know the drill.

To obtain logs and information about crashes in a production setup, make sure that you have access to an external Sentry instance (e.g. by creating an account with, and set the SENTRY_DSN variable. Logs of level logging.ERROR are sent as Sentry events. Therefore, in order to send a Sentry event use:

import logging
logging.error("This event is sent to Sentry", extra={"<example_key>": "<example_value>"})

The extra parameter allows you to send additional information about the context of this error.

You will probably also need to setup the Mail backend, for example by adding a Mailgun API key and a Mailgun sender domain, otherwise, the account creation view will crash and result in a 500 error when the backend attempts to send an email to the account owner.


If your email server used to send email isn’t configured properly (Mailgun by default), attempting to send an email will cause an Internal Server Error.

By default, django-allauth is setup to have emails verifications mandatory, which means it’ll send a verification email when an unverified user tries to log-in or when someone tries to sign-up.

This may happen just after you’ve setup your Mailgun account, which is running in a sandbox subdomain by default. Either add your email to the list of authorized recipients or verify your domain.

Optional: Use AWS IAM Role for EC2 instance

If you are deploying to AWS, you can use the IAM role to substitute AWS credentials, after which it’s safe to remove the AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY from .envs/.production/.django. To do it, create an IAM role and attach it to the existing EC2 instance or create a new EC2 instance with that role. The role should assume, at minimum, the AmazonS3FullAccess permission.

HTTPS is On by Default

SSL (Secure Sockets Layer) is a standard security technology for establishing an encrypted link between a server and a client, typically in this case, a web server (website) and a browser. Not having HTTPS means that malicious network users can sniff authentication credentials between your website and end users’ browser.

It is always better to deploy a site behind HTTPS and will become crucial as the web services extend to the IoT (Internet of Things). For this reason, we have set up a number of security defaults to help make your website secure:

  • If you are not using a subdomain of the domain name set in the project, then remember to put your staging/production IP address in the DJANGO_ALLOWED_HOSTS environment variable (see Settings) before you deploy your website. Failure to do this will mean you will not have access to your website through the HTTP protocol.
  • Access to the Django admin is set up by default to require HTTPS in production or once live.

The Traefik reverse proxy used in the default configuration will get you a valid certificate from Lets Encrypt and update it automatically. All you need to do to enable this is to make sure that your DNS records are pointing to the server Traefik runs on.

You can read more about this feature and how to configure it, at Automatic HTTPS in the Traefik docs.

(Optional) Postgres Data Volume Modifications

Postgres is saving its database files to the production_postgres_data volume by default. Change that if you want something else and make sure to make backups since this is not done automatically.

Building & Running Production Stack

You will need to build the stack first. To do that, run:

docker-compose -f production.yml build

Once this is ready, you can run it with:

docker-compose -f production.yml up

To run the stack and detach the containers, run:

docker-compose -f production.yml up -d

To run a migration, open up a second terminal and run:

docker-compose -f production.yml run --rm django python migrate

To create a superuser, run:

docker-compose -f production.yml run --rm django python createsuperuser

If you need a shell, run:

docker-compose -f production.yml run --rm django python shell

To check the logs out, run:

docker-compose -f production.yml logs

If you want to scale your application, run:

docker-compose -f production.yml up --scale django=4
docker-compose -f production.yml up --scale celeryworker=2


don’t try to scale postgres, celerybeat, or traefik.

To see how your containers are doing run:

docker-compose -f production.yml ps

Example: Supervisor

Once you are ready with your initial setup, you want to make sure that your application is run by a process manager to survive reboots and auto restarts in case of an error. You can use the process manager you are most familiar with. All it needs to do is to run docker-compose -f production.yml up in your projects root directory.

If you are using supervisor, you can use this file as a starting point:

command=docker-compose -f production.yml up

Move it to /etc/supervisor/conf.d/{{cookiecutter.project_slug}}.conf and run:

supervisorctl reread
supervisorctl update
supervisorctl start {{cookiecutter.project_slug}}

For status check, run:

supervisorctl status

Docker Security

This section contains a list of security issues identified by the Docker Bench for Security tool, after a deployment in an Ubuntu 16.04 machine using the instructions in Building & Running Production Stack, and possible fixes.


After applying some of the fixes you might need to rebuild the stack, otherwise the issues might still be reported when re-running Docker Bench for Security.


The numbers in the headings correspond to the Docker Bench for Security test number.

1.2.1 - Ensure a separate partition for containers has been created

1.2.3 - Ensure auditing is configured for the Docker daemon and files/directories

Install auditd:

$ sudo apt-get install auditd

Edit the auditing system rules:

$ sudo vim /etc/audit/audit.rules

These rules instruct auditd to watch (-w) the specified file or directory and log any writes or attribute changes (-p wa) to those files:

-w /etc/default/docker -p wa
-w /etc/docker -p wa
-w /etc/docker/daemon.json -p wa
-w /etc/sysconfig/docker -p wa
-w /lib/systemd/system/docker.service -p wa
-w /lib/systemd/system/docker.socket -p wa
-w /project/docker -p wa
-w /usr/bin/docker -p wa
-w /usr/bin/containerd -p wa
-w /usr/bin/runc -p wa
-w /var/lib/docker -p wa

Restart autditd:

$ sudo systemctl restart auditd

2 - Docker daemon configuration

    "data-root": "/project/docker",
    "icc": false,
    "live-restore": true,
    "log-driver": "syslog",
    "no-new-privileges": true,
    "userland-proxy": false,
    "userns-remap": "default"

For more information on how to configure the Docker daemon see the official Docker daemon documentation. Below is a short explanation for each of the configuration options in daemon.json.

Root directory of persistent Docker state (default “/var/lib/docker”)
2.1 - Ensure network traffic is restricted between containers on the default bridge
2.13 - Ensure live restore is Enabled
2.12 - Ensure centralized and remote logging is configured
2.18 - Ensure containers are restricted from acquiring new privileges
2.15 - Ensure Userland Proxy is Disabled
2.8 - Enable user namespace support

4.5 - Ensure Content trust for Docker is Enabled

To enable content trust for all users and sessions:

$ echo "DOCKER_CONTENT_TRUST=1" | sudo tee -a /etc/environment

For more information see the Docker content trust documentation.

4.6 - Ensure that HEALTHCHECK instructions have been added to container images

This should also cover the issue with 5.26 - Ensure that container health is checked at runtime.

For example, to check every five minutes or so that a web-server is able to serve the site’s main page within three seconds:

HEALTHCHECK --interval=5m --timeout=3s \
CMD curl -f http://localhost/ || exit 1

5.2 - Ensure that, if applicable, SELinux security options are set

5.7 - Ensure privileged ports are not mapped within containers

Mapping http port 80 and https port 443 is necessary for traefik/webserver. All the other ports in the stack are not privileged ports.

5.10 - Ensure that the memory usage for containers is limited

Runtime options with Memory.

5.11 - Ensure CPU priority is set appropriately on the container

Runtime options with CPUs.

5.12 - Ensure that the container’s root filesystem is mounted as read only

Mount host-sensitive directories as read-only. In the default cookiecutter configuration no host-sensitive directories are shared with the containers.

5.13 - Ensure that incoming container traffic is bound to a specific host interface

5.14 - Ensure that the ‘on-failure’ container restart policy is set to ‘5’

Restart policy.

5.25 - Ensure that the container is restricted from acquiring additional privileges

Set in /etc/docker/daemon.json.

5.27 - Ensure that Docker commands always make use of the latest version of their image

5.28 - Ensure that the PIDs cgroup limit is used