My understanding is if you want to access the app on port 3000, you would use `docker run -d -p 3000:8080 ...` for a Docker command. For Docker Compose, it's: ``` nginx_recipes: ports: - 3000:80 ```
11 KiB
!!! success "Recommended Installation" Setting up this application using Docker is recommended. This does not mean that other options are bad, just that support is much easier for this setup.
It is possible to install this application using many Docker configurations.
Please read the instructions/notes on each example carefully and decide if this is the way for you.
Docker
The docker image (vabene1111/recipes
) simply exposes the application on the container's port 8080
.
It can be run and accessed on port 80 using:
docker run -d \
-v ./staticfiles:/opt/recipes/staticfiles \
-v ./mediafiles:/opt/recipes/mediafiles \
-p 80:8080 \
-e SECRET_KEY=YOUR_SECRET_KEY \
-e DB_ENGINE=django.db.backends.postgresql \
-e POSTGRES_HOST=db_recipes \
-e POSTGRES_PORT=5432 \
-e POSTGRES_USER=djangodb \
-e POSTGRES_PASSWORD=YOUR_POSTGRES_SECRET_KEY \
-e POSTGRES_DB=djangodb \
--name recipes_1 \
vabene1111/recipes
Please make sure, if you run your image this way, to consult the .env.template file in the GitHub repository to verify if additional environment variables are required for your setup.
Docker Compose
The main, and also recommended, installation option is to install this application using Docker Compose.
- Choose your
docker-compose.yml
from the examples below. - Download the
.env
configuration file withwget
, then edit it accordingly.wget https://raw.githubusercontent.com/vabene1111/recipes/develop/.env.template -O .env
- Start your container using
docker-compose up -d
.
Plain
This configuration exposes the application through an nginx web server on port 80 of your machine.
version: "3"
services:
db_recipes:
restart: always
image: postgres:11-alpine
volumes:
- ./postgresql:/var/lib/postgresql/data
env_file:
- ./.env
web_recipes:
image: vabene1111/recipes
restart: always
env_file:
- ./.env
volumes:
- staticfiles:/opt/recipes/staticfiles
- mediafiles:/opt/recipes/mediafiles
- nginx_config:/opt/recipes/nginx/conf.d
depends_on:
- db_recipes
nginx_recipes:
image: nginx:mainline-alpine
restart: always
ports:
- 80:80
env_file:
- ./.env
volumes:
- nginx_config:/etc/nginx/conf.d:ro
- staticfiles:/static
- mediafiles:/media
volumes:
nginx_config:
staticfiles:
mediafiles:
driver: local
driver_opts:
type: 'none'
o: 'bind'
device: './mediafiles'
Reverse Proxy
Most deployments will likely use a reverse proxy.
Traefik
If you use traefik, this configuration is the one for you.
!!! info Traefik can be a little confusing to setup. Please refer to their excellent documentation. If that does not help, this little example might be for you.
version: "3"
services:
db_recipes:
restart: always
image: postgres:11-alpine
volumes:
- ./postgresql:/var/lib/postgresql/data
env_file:
- ./.env
networks:
- default
web_recipes:
image: vabene1111/recipes
restart: always
env_file:
- ./.env
volumes:
- staticfiles:/opt/recipes/staticfiles
- mediafiles:/opt/recipes/mediafiles
- nginx_config:/opt/recipes/nginx/conf.d
depends_on:
- db_recipes
networks:
- default
nginx_recipes:
image: nginx:mainline-alpine
restart: always
env_file:
- ./.env
volumes:
- nginx_config:/etc/nginx/conf.d:ro
- staticfiles:/static
- mediafiles:/media
labels: # traefik example labels
- "traefik.enable=true"
- "traefik.http.routers.recipes.rule=Host(`recipes.mydomain.com`, `recipes.myotherdomain.com`)"
- "traefik.http.routers.recipes.entrypoints=web_secure" # your https endpoint
- "traefik.http.routers.recipes.tls.certresolver=le_resolver" # your cert resolver
networks:
- default
- traefik
networks:
default:
traefik: # This is you external traefik network
external: true
volumes:
nginx
staticfiles
mediafiles:
driver: local
driver_opts:
type: 'none'
o: 'bind'
device: './mediafiles'
nginx-proxy
This is a docker compose example using jwilder's nginx reverse proxy in combination with jrcs's letsencrypt companion.
Please refer to the appropriate documentation on how to setup the reverse proxy and networks.
Remember to add the appropriate environment variables to .env
file:
VIRTUAL_HOST=
LETSENCRYPT_HOST=
LETSENCRYPT_EMAIL=
version: "3"
services:
db_recipes:
restart: always
image: postgres:11-alpine
volumes:
- ./postgresql:/var/lib/postgresql/data
env_file:
- ./.env
networks:
- default
web_recipes:
image: vabene1111/recipes
restart: always
env_file:
- ./.env
volumes:
- staticfiles:/opt/recipes/staticfiles
- mediafiles:/opt/recipes/mediafiles
- nginx_config:/opt/recipes/nginx/conf.d
depends_on:
- db_recipes
networks:
- default
nginx_recipes:
image: nginx:mainline-alpine
restart: always
env_file:
- ./.env
volumes:
- nginx_config:/etc/nginx/conf.d:ro
- staticfiles:/static
- mediafiles:/media
networks:
- default
- nginx-proxy
networks:
default:
nginx-proxy:
external:
name: nginx-proxy
volumes:
nginx
staticfiles
mediafiles:
driver: local
driver_opts:
type: 'none'
o: 'bind'
device: './mediafiles'
Additional Information
Nginx vs Gunicorn
All examples use an additional nginx
container to serve mediafiles and act as the forward facing webserver.
This is technically not required but very much recommended.
I do not 100% understand the deep technical details but the developers of gunicorn, the WSGi server that handles the Python execution, explicitly state that it is not recommended to deploy without nginx. You will also likely not see any decrease in performance or a lot of space used as nginx is a very light container.
!!! info Even if you run behind a reverse proxy as described above, using an additional nginx container is the recommended option.
If you run a small private deployment and don't care about performance, security and whatever else feel free to run without a ngix container.
!!! warning
When running without nginx make sure to enable GUNICORN_MEDIA
in the .env
. Without it, media files will be uploaded
but not shown on the page.
For additional information please refer to the 0.9.0 Release and Issue 201 where these topics have been discussed. See also refer to the official gunicorn docs.
Nginx Config
In order to give the user (you) the greatest amount of freedom when choosing how to deploy this application the webserver is not directly bundled with the Docker image.
This has the downside that it is difficult to supply the configuration to the webserver (e.g. nginx). Up until
version 0.13.0
, this had to be done manually by downloading the nginx config file and placing it in a directory that
was then mounted into the nginx container.
From version 0.13.0
, the config file is supplied using the application image (vabene1111/recipes
). It is then mounted
to the host system and from there into the nginx container.
This is not really a clean solution, but I could not find any better alternative that provided the same amount of usability. If you know of any better way, feel free to open an issue.
Using Proxy Authentication
!!! Info "Community Contributed Tutorial" This tutorial was provided by a community member. Since I do not use reverse proxy authentication, I cannot provide any assistance should you choose to use this authentication method.
In order use proxy authentication you will need to:
- Set
REVERSE_PROXY_AUTH=1
in the.env
file - Update your nginx configuration file
Using any of the examples above will automatically generate a configuration file inside a docker volume.
Use docker volume inspect recipes_nginx
to find out where your volume is stored.
!!! warning "Configuration File Volume"
The nginx config volume is generated when the container is first run. You can change the volume to a bind mount in the
warning docker-compose.yml
, but then you will need to manually create it. See section Volumes vs Bind Mounts
below
for more information.
The following example shows a configuration for Authelia:
server {
listen 80;
server_name localhost;
client_max_body_size 16M;
# serve static files
location /static/ {
alias /static/;
}
# serve media files
location /media/ {
alias /media/;
}
# Authelia endpoint for authentication requests
include /config/nginx/auth.conf;
# pass requests for dynamic content to gunicorn
location / {
proxy_set_header Host $host;
proxy_pass http://web_recipes:8080;
# Ensure Authelia is specifically required for this endpoint
# This line is important as it will return a 401 error if the user doesn't have access
include /config/nginx/authelia.conf;
auth_request_set $user $upstream_http_remote_user;
proxy_set_header REMOTE-USER $user;
}
# Required to allow user to logout of authentication from within Recipes
# Ensure the <auth_endpoint> below is changed to actual the authentication url
location /accounts/logout/ {
return 301 http://<auth_endpoint>/logout
}
}
Please refer to the appropriate documentation on how to setup the reverse proxy, authentication, and networks.
Ensure users have been configured for Authelia, and that the endpoint recipes is pointed to is protected but available.
There is a good guide to the other additional files that need to be added to your nginx set up at the Authelia Docs.
Remember to add the appropriate environment variables to .env
file (example for nginx proxy):
VIRTUAL_HOST=
LETSENCRYPT_HOST=
LETSENCRYPT_EMAIL=
PROXY_HEADER=
Volumes vs Bind Mounts
Since I personally prefer to have my data where my docker-compose.yml
resides, bind mounts are used in the example
configuration files for all user generated data (e.g. Postgresql and media files).
Please note that there is a difference in functionality between the two and you cannot always simply interchange them.
You can move everything to volumes if you prefer it this way, but you cannot convert the nginx config file to a bind mount. If you do so you will have to manually create the nginx config file and restart the container once after creating it.