Lynchburg Server Dockerfile

Project Lynchburg
https://git.noc.ruhr-uni-bochum.de/lynchburg/lynchburg-docker

Introduction

This Dockerfile can be used to easily install the Lynchburg Server web2py application. You can use the Docker container with or without a reverse proxy, but using one is recommended.

License

The Dockerfile and helper scripts are put into the public domain. For more information check the LICENSE file.
For information about contributors goto the Project Lynchburg landing page

NOTE: The temporary workaround file websocket_messaging.py is still licensed under LGPLv3 in compliance with the web2py LICENSE.

Installation

1. Build the docker image using docker build --build-arg "ADMIN_PASSWORD=<YOUR_ADMIN_PASSWORD>" -t lynchburg ..
   • Additionally add another --build-arg "CONTACT=<CONTACT LINK>", if you wish to apply a link to your contact page to the contact button. By default the link is set to #. Note: Adding a contact page is a legal requirement in some jurisdictions.
   • Additionally add another --build-arg "REVERSE_PROXY=false", if you wish to use the docker container directly without a reverse proxy (not recommended). Note that using HTTPS for Lynchburg will require a lot of effort when not using a reverse proxy.
   • Additionally add another --build-arg "DB_URI=<URI>", if you wish to use a real database instead of the default sqlite one. Valid <URI> strings can be found here.
2. Create a new docker container using docker create --name lynchburg lynchburg
3. Start the container using docker container start lynchburg

Reverse proxy

The docker container exposes port 80 as uswgi and 8888 as WebSocket server. Use a reverse proxy for BOTH ports. Either forward the container ports to some host port by using the -p flag when creating the container, or use docker network inspect bridge to get the container specific IP.

If you are using nginx the basic configuration should look like this:

server {
    listen 0.0.0.0:80;

    # This is the websocket reverse proxy. The /realtime/ part of the URL must be forwarded too.
    # The Host must be changed to 127.0.0.1:8888
    # Replace the <IP> in proxy_pass with the container specific IP, or use a host ip:port if you used the -p flag when creating the container
    location /realtime/ {
        proxy_pass                  http://<IP>:8888;
        proxy_http_version          1.1;
        proxy_set_header Upgrade    $http_upgrade;
        proxy_set_header Connection $connection_upgrade;
        proxy_set_header Host       127.0.0.1:8888;
        error_log                   /var/log/nginx/lynchburg_ws.err warn;
        access_log                  /var/log/nginx/lynchburg_ws.log;
    }

    # This is the http reverse proxy
    # Replace the <IP> in proxy_pass with the container specific IP, or use a host ip:port if you used the -p flag when creating the container
    location / {
        uwsgi_pass      <IP>:80;
        include         uwsgi_params;
        uwsgi_param     UWSGI_SCHEME $scheme;
        uwsgi_param     SERVER_SOFTWARE nginx/$nginx_version;
        error_log       /var/log/nginx/lynchburg.err warn;
        access_log      /var/log/nginx/lynchburg.log;
    }
}

HTTPS should work flawlessly when using a reverse proxy. Just configure the webserver accordingly.

Connection Upgrade

nginx, and possibly other webservers, require a connection upgrade when using WebSockets. For nginx follow the standard practise of defining $http_upgrade and $connection_upgrade as:

map $http_upgrade $connection_upgrade {  
    default upgrade;
    ''      close;
}

Put that map block into the http block of your nginx configuration. The default file path for the nginx config is /etc/nginx/nginx.conf