All Projects → douglasmiranda → ddpt

douglasmiranda / ddpt

Licence: MIT license
Template for Django Projects - From development to production with Docker Swarm.

Programming Languages

python
139335 projects - #7 most used programming language
Makefile
30231 projects
HTML
75241 projects
Dockerfile
14818 projects
HCL
1544 projects
shell
77523 projects

Projects that are alternatives of or similar to ddpt

Wemake Django Template
Bleeding edge django template focused on code quality and security.
Stars: ✭ 1,141 (+2012.96%)
Mutual labels:  cookiecutter, caddy
docker-volume-hetzner
Docker Volume Plugin for accessing Hetzner Cloud Volumes
Stars: ✭ 81 (+50%)
Mutual labels:  docker-swarm
Ansible Dockerswarm
Docker Engine clustering using "Swarm Mode" and Ansible
Stars: ✭ 226 (+318.52%)
Mutual labels:  docker-swarm
docker-cheat-sheet
Important Docker Commands Ordered by Compute (Services), Network, Storage then by Docker CLI, Dockerfile, Compose, and Swarm
Stars: ✭ 21 (-61.11%)
Mutual labels:  docker-swarm
Fabricio
Docker deploy automation tool
Stars: ✭ 250 (+362.96%)
Mutual labels:  docker-swarm
FastAPI Tortoise template
FastAPI - Tortoise ORM - Celery - Docker template
Stars: ✭ 144 (+166.67%)
Mutual labels:  docker-swarm
Docker Prometheus Swarm
Sample prometheus that can be used as a sample to get Swarm cluster metrics
Stars: ✭ 210 (+288.89%)
Mutual labels:  docker-swarm
vagrant-boot2docker-swarm
A multi-machine Docker swarm Vagrant environment
Stars: ✭ 16 (-70.37%)
Mutual labels:  docker-swarm
docker-roj
Portable environments for Docker swarms and stacks
Stars: ✭ 31 (-42.59%)
Mutual labels:  docker-swarm
Labs
Container Orchestration Labs
Stars: ✭ 14 (-74.07%)
Mutual labels:  docker-swarm
jupyterlab-starters
Starter notebooks and directories in JupyterLab
Stars: ✭ 32 (-40.74%)
Mutual labels:  cookiecutter
docker-ssl-reverse-proxy
Easy-to-use auto-SSL reverse proxy as a Docker container based on Caddy and Let’s Encrypt
Stars: ✭ 22 (-59.26%)
Mutual labels:  caddy
python-template
Netherlands eScience Center Python Template
Stars: ✭ 136 (+151.85%)
Mutual labels:  cookiecutter
Docker Mastery For Nodejs
Docker Mastery for Node.js Projects, From a Docker Captain
Stars: ✭ 231 (+327.78%)
Mutual labels:  docker-swarm
retrocookie
Retrocookie updates Cookiecutter templates with changes from generated projects
Stars: ✭ 22 (-59.26%)
Mutual labels:  cookiecutter
Docker Traefik Prometheus
A Docker Swarm Stack for monitoring Traefik with Promethues and Grafana
Stars: ✭ 215 (+298.15%)
Mutual labels:  docker-swarm
caddy-tlsconsul
🔒 Consul K/V storage for Caddy Web Server / Certmagic TLS data
Stars: ✭ 89 (+64.81%)
Mutual labels:  caddy
Docker-Swarm-DDOS
How to create a Denial of Service Attack through Docker Swarm.
Stars: ✭ 36 (-33.33%)
Mutual labels:  docker-swarm
cookiecutter-esp32-webserver
Cookiecutter template to get you quickly started with an ESP32-based webserver project.
Stars: ✭ 13 (-75.93%)
Mutual labels:  cookiecutter
FastAPI-template
Feature rich robust FastAPI template.
Stars: ✭ 660 (+1122.22%)
Mutual labels:  cookiecutter

Django Project Template

Template for Django Projects - From development with docker-compose to production with Docker Swarm (docker stack deploy).

Features

  • Django 2.2 + Python 3.6 / 3.7
  • Using Alpine Linux based images
  • Setup close to production + nice dev tools when local
  • Ready to use Docker Secrets for keeping your sensitive data safe
  • Ready to use Docker Configs if you want
  • Development:
    • Build and run with Docker Compose
    • Django Extensions
    • Django Debug Toolbar
    • ipdb
    • Format code style with Black!
  • Production:
    • Deploy with Docker Swarm (docker stack command)
    • Run your Django using gunicorn
    • Reverse proxy with Caddy (easy HTTPS)
    • Serve static files with Whitenoise (planning to remove, not ideal for replicated Django)
    • Redis for Cache and Session Storage
    • Send emails via Anymail (using Mailgun by default, but switchable)
  • Bonus: Check the Makefile, you'll see lots of shortcuts to speed up your development

TODO

Optional Integrations

These features can be enabled during initial project setup.

  • (Development) Integration with MailHog for local email testing
  • (Production) Integration with Mailgun for local email testing
  • (Production) Integration with Sentry for error logging
  • (Production) Media storage using Amazon S3
  • Django Admin theme

Usage - Development

Introducing Cookiecutter

Let's pretend you want to create a Django project called "awesome_project". Rather than using startproject and then editing the results to include your name, email, and various configuration issues that always get forgotten until the worst possible moment, get cookiecutter to do all the work.

First, get Cookiecutter. Trust me, it's awesome:

pip install cookiecutter

Start the project

Now run it against this repo:

cookiecutter https://github.com/douglasmiranda/ddpt

You'll be prompted for some values. Provide them, then a Django project will be created for you.

Answer the prompts with your own desired options. For example:

$ cookiecutter https://github.com/douglasmiranda/ddpt
project_name [Project Name]: Awesome Project
django_project_name [awesome_project]: awesome_project
author_name [Your Name or Company Name]: Awesome Company
email [[email protected]]: [email protected]
description [A short description of the project.]: An Awesome Project!
domain_name [example.com]: awesomeproject.dev
version [0.1.0]: 0.1.0
timezone [UTC]: UTC
language_code [en-us]: en-us
use_admin_theme [y]: y
use_mailgun [n]: y
use_mailhog [y]: y
use_s3 [n]: y
use_sentry_for_error_reporting [n]: y

Enter the project and take a look around:

~/dev
❯ cd awesome_project/

~/dev/awesome_project
❯ ls
awesome_project/  deployment/  docker-compose.yml  Dockerfile*  Makefile  manage.py*  README.md*  requirements/

Let's see the project structure:

❯ tree
.
├── awesome_project
│   ├── config
│   │   ├── base.py
│   │   ├── __init__.py
│   │   ├── local.py
│   │   ├── production.py
│   │   ├── test.py
│   │   ├── urls.py
│   │   └── wsgi.py
│   ├── core
│   │   ├── admin.py
│   │   ├── apps.py
│   │   ├── __init__.py
│   │   ├── migrations
│   │   │   └── __init__.py
│   │   ├── models.py
│   │   ├── templates
│   │   │   └── core
│   │   │       └── home.html
│   │   ├── urls.py
│   │   └── views.py
│   ├── __init__.py
│   └── users
│       ├── admin.py
│       ├── apps.py
│       ├── __init__.py
│       ├── migrations
│       │   ├── 0001_initial.py
│       │   └── __init__.py
│       ├── models.py
│       ├── urls.py
│       └── views.py
├── deployment
│   ├── caddy
│   │   ├── Caddyfile
│   │   └── Dockerfile
│   ├── db-migration.yml
│   ├── django
│   │   ├── django-entrypoint.sh
│   │   └── django-health check.sh
│   ├── docker-stack.caddy.yml
│   ├── docker-stack.django.yml
│   ├── docker-stack.postgres.yml
│   ├── docker-stack.redis.yml
│   ├── Makefile
│   └── README.md
├── docker-compose.yml
├── Dockerfile
├── Makefile
├── manage.py
├── README.md
└── requirements
    ├── base.txt
    ├── dev.txt
    └── test.txt

Running your newly created Django project

If you know the basics of Docker, Docker Compose and the Django management commands (python manage.py / django-admin), you all set. But I do have one thing or two to show you in the next topics.

How to run

Note about building the Django image

I'm using a user called django for this container, for production and development. This user inside the container has to match the UID of the host user so you don't have those permissions issues when sharing a volume between the container and your host OS. More on this subject here.

When I'm using MAC this doesn't matter (because of how Docker for MAC works) and when I'm in Linux, my UID is probably the default 1000 (find with id -u $USER), so I think it's a good idea to set this as default. But since it's an ARG you can customize the build, with build args in Docker. With Docker Compose, for example:

# Let's find what's our UID
$ id -u $USER
1001
# If it's 1001:
docker-compose build --build-arg DJANGO_USER_UID=1001

Build, run and manage with Makefile

There are many commands we run all day long during development, so I made a Makefile that contains some good common tasks to help you get more productive. Next, we'll see what's in that Makefile.

Makefile

Don't fear the Makefile, it's not that weird to read. It contains common commands that we run during development, so you can change and improve to fit your daily flow.

If you run make help you'll see what's available for you to use.

command description
help show make targets
build build necessary stuff for our project to run (docker images)
run start containers with docker-compose and attach to logs
rund start containers with docker-compose (detached mode)
stop stop all running containers for this project
enter enter the Django container (want to play freely with manage.py commands? just make enter and have fun)
enter-root enter the Django container as ROOT
shell enter the Django container and run the shell_plus or shell
test enter the running Django container and run tests with pytest
debug after you set the breakpoint() in your code, attach to the Django container
compilemessages Localization: enter the running Django container and compile .po files (create .mo files)
format format code using Black
format-diff see how your code will look like after being formatted with Black
logs attach to logs of containers for this project
list-packages list installed Python packages with pipdeptree or pip freeze. If pipdeptree is available it list possible dependencies conflicts
list-outdated list outdated Python packages
py-info useful info about your Python Environment (Container)
clean-py clean python artifacts (cache, tests, build stuff...)
flush-cache Clear cache, in our case Redis. Using "default" connection. See CACHES in settings.py
remove stop running containers and remove them
remove-full DANGER stop and remove running containers, VOLUMES and NETWORKS associated to them
first-run DANGER This will run migrations and create a INSECURE superuser that you should only use in dev environment
fix-volume-permission Fix common volume permissions
1 - Build
make build

Our build target it's basically docker-compose build. But it will discover our (host) user UID and set as the UID of the django user inside the container.

Why create a "django" user in the Django image?

2 - Run
# For run attached
make run
# Run detached
make rund

Basically docker-compose up. But with the --no-build flag to make sure you run make build which takes care of the user UID stuff.

3 - Firt things on first run

Now you can just enter the Django container:

# Using our shortcut
make enter
# Or with docker-compose
docker-compose exec django sh

And with python manage.py commands you do what you usually do. (Run migrations, create super user...)

But I do have a shortcut for this too. xD

# Using our shortcut
make first-run

Basically it runs migrations and add a superuser with username a and password a.

If for some reason you still have a problem with volume permissions, there's a fix in the Makefile too, just run make fix-volume-permission.

Usage - Deployment/Production

I'm still working in the Deployment/Production docs, but you can check at deployment/.

My plan is to briefly talk and/or give examples about Provisioning (Terraform), Configuration Management (Ansible), Object Storage, Docker and more.

Why Am I doing things this way?

Well, this is an opinionated project template, if you have more questions, don't be shy, open an issue.

Next, I'll try to answer some questions.

Why create a "django" user in the Django image?

Unprivileged user

That's because it's more secure to run processes with non-privileged users. You could choose to run in development as root, but in Django we're always running commands that generate files, and you'll be doing that from inside the container, therefore creating files as root and you can't write to them with your normal host user. You'll be always executing sudo chown .... Plus I'm trying to keep things close to production here.

In Linux what matters is not the human-readable user name like "django". What really matters is the UID, and they can differ from one Linux distribution to another, so you have to make sure to use the same UID for both the container user and you host user.

You can find your user UID with id -u $USER.

Fix permissions in a shared code volume using a common UID

As you can see in our docker-compose.yml file we have our Django code shared with the Django container. In projects like this, we create files from outside and from inside the container, and this can lead to some permission issues in development. (Not only in dev but for our production version we apply the correct owner to the files when copying, so no problem there.)

The approach I chose here was to use your user UID as the UID of the user django inside the container.

If you're in Docker for MAC, you won't face that issue because it works a litte different when sharing volumes from host to user.

Why Alpine and not Debian or Ubuntu?

Alpine Linux is great, sometimes you'll need to build something from source, discover what the proper package name and things like that, but it's more secure and lightweight, do some search and you may like.

Debian, Ubuntu and other images are great too, you can make changes and use them. Just make sure to use the official images.

Where's NGINX?

I'm using Caddy as a reverse proxy for Django, it's simple, great, full of features, easy HTTPS. Depending on the size of your project and audience, you'll probably face some other issues with optimization before you get to optimize your Web Server.

*But I will provide NGINX configs with HTTPS soon.

Why Docker Swarm and not Docker Compose in production?

Well, you can, and maybe you should.

But let me try to convince you to use Docker Swarm even if you're running in a single node/host.

(I'll try to give some examples of practical things you get with Docker Swarm, for more detailed, advanced stuff you should really do some research on the subject.)

It will be familiar

You can deploy in Docker Swarm with the docker stack command almost the same docker-compose.yml file you're used to. There are some things you can use in one and not the other, but you'll be fine, nothing impossible to fix.

Advanced Deployment

Control

You can control resources usage, how it will be restarted if needed, health checks, replicas and more.

Automatic rollback

The thing about health check is that you can do auto-rollback, which is awesome. Every time you redeploy, if your new update breaks the health check, Docker will automatically rollback to the previous configuration.

Zero downtime deployment

If you have control of replicas, how they are started, restarted, updated and can check the health of you containers, you can enjoy zero downtime deployments.

The flow is basically:

  • You make a new deployment
  • With docker stack deploy, the replicas will start the update process
    • you can tell Docker to start the new replicas
    • make sure they are healthy
    • stop the old replicas
  • If everything went ok, docker ingress network will redirect the requests to the new replicas
  • If the updated replicas are not fine, the health check reported failure, Docker start the rollback to the previous settings

Of course, it can be more complicated than that, you probably need to run some simulated scenarios in order to decide if this is enough for your project.

Security

Learn about Docker Secrets and stop using plain environment variables for sensitive data.

You may want to learn about connecting to Docker remotely using TLS and improve your deployment. Do not expose Docker without configuring TLS!

Some Docker Swarm notes

Single Node

As I said before, you can enjoy Docker Swarm in a single machine, you don't need to configure a multi-node setup.

If you are in a single node, you can use volumes and don't even need to worry about replicas placement, almost just like with Docker Compose.

Multiple Nodes

Well, I can't possibly cover here in some notes the challenges that distributed systems bring. But I can list some specific notes to Django in a multi-node setup and Docker Swarm.

Databases and Storages

I won't cover clustering databases, that's really extensive.

I won't cover distributed storages either for the same reason.

But let's say your scenario is: You have to go multi-node, but only for some workers and some containers you're running, maybe it's because they are memory/CPU intensive or something like that. Your Django application, some volume, maybe the database, can stay in the same node(s) and you would like a nice solution to fit that design.

You could get something from placement in your Docker Stack/Compose files.

"Easy" solution is: Use something like S3 for storage and any database as a service from your hosting/cloud provider and go crazy on replicate and manage your application/worker containers with Docker Swarm, that way you probably will face fewer issues.


Submit a Pull Request

Keep your pull requests small. Better yet, explain what you plan to do in the issues, it will save time in the end.

TODO

  • Deployment docs #6
  • Deployment: Provisioning (info, examples, docs...) #2
  • More info about Docker Swarm
    • maybe more information about using even if it's single node
  • More information on Storages (Static/Media files) #7
    • remove / make optional whitenoise
    • alternative to S3 (maybe Minio would be nice)
    • CDN
  • CI/CD docs and/or examples #3

Related links

Docker

Python/Django

Docker Images

Others

Me

Note that the project description data, including the texts, logos, images, and/or trademarks, for each open source project belongs to its rightful owner. If you wish to add or remove any projects, please contact us at [email protected].