This is a very basic configuration, while I will likely work on improving it, it most likely is not a very secure setup, so I do not recommend using this for hosting sensitive information.
- Docker/Nginx/DigitalOcean Web Server
While there are likely more straightforward approaches when it comes to manually setting up some hosting configuration, this was personally a nice exploration into some of the more manual aspects of hosting and deployment, and I hope it helps anyone else aswell. Additionally its nice to have a reference for some of the fundamentals.
This is a basic example of a simple server making use of various technologies such as:
nginx: Web server and reverse proxyDigitalOcean Droplet: Cloud VM which is running the serverdocker-compose: To orchestrate thenginxandexpressserver containerscertbot: To get the SSL certificatesworkflow: To automatically deploy the server via SSH to the VM
So there are various aspects to this, but we can break it down into parts.
The core aspect here is the Nginx server. We created a custom config to run a basic server which listens on all network interfaces on port 80. This configuration is defined in default.conf. This configuration is then copied to the /etc/nginx/conf.d/ folder. After this, when the server runs, it includes this configuration in its main nginx.conf file.
The Nginx Docker image by default looks in /usr/share/nginx/html for content, so we copy over the contents of the data folder here.
We then define our custom index.html page in the Nginx config with the following location block, which is then accessible via the URL asdfxyz.xyz
location / {
root /usr/share/nginx/html/html;
index index.html index.htm;
}
For the image, we define the location using the following block
location /image/ {
root /usr/share/nginx/html/;
}
The image is then accessed via the URL asdfxyz.xyz/image/_33A7500.JPG
To test out the reverse proxy functionality, that is, on a specific (or all) routes, we redirect the request somewhere else, such as another server running on the same machine.
Doing so is rather simple. We first set up a basic express server in backend/app.js
const express = require('express');
const app = express();
const port = 3000;
// Route to serve the HTML page
app.get('/', (req, res) => {
res.send('<h1>Hello, World!</h1>');
});
// Start the server
app.listen(port, () => {
console.log(`Server is running`);
});Keeping in line with the Docker philosophy, we create another container for the server which just exposes the port it runs on and then runs the server.
Now, since we have two Dockerfiles, one to run the nginx server in the root directory and one to run the node server in the backend directory, we preferably need some way to build and run both of them easily. This is where docker-compose comes in handy; it's a nice way to manage multiple containers via a simple .yml file.
Ignoring the certbot stuff for the time being, you should see two containers: backend, which is running the node server, and nginx, which is running the Nginx server. Both containers perform a 1:1 mapping of the VM port to the containers exposed port. Aside from that, they just use what's defined in the Dockerfile to build.
To then build and run the containers, we use the command docker-compose up -d --build
-d: if we want it to run in the background
And to shut the containers down, we use docker-compose down
Here is where the spicy part comes in. If we, for example, want the route asdfxyz.xyz/api/ to access content being served by the node server, what we can do is use proxy_pass in a location block to redirect the request to that of the server. Which is exactly what is happening here.
location /api/ {
# backend is the name of the container, don't use `localhost`
proxy_pass http://backend:3000/;
...
}
I would recommend you first test acquiring the certificats using the --dry-run flag to make sure yoru configuration is setup correctly, because you can get locked out of acquiring certificates if something fails too many times.
docker compose run --rm certbot certonly --webroot --webroot-path /var/www/certbot/ -d example.orgFinally, we get the actual certificates for the domain. For this I made a bash script which you can run with chmod +x setup_ssl.sh && ./setup_ssl.sh. You can also just pass the volumes in the command directly, to do so alter the script below based on the official documentation, but in either case the container places the certificate files in the specified volumes on your system.
#!/bin/bash
email_arg="--email [email protected]"
domain_args="-d example.com"
docker compose run --rm --entrypoint "\
certbot certonly --webroot -w /var/www/certbot \
$email_arg \
$domain_args \
--agree-tos \
--force-renewal" certbot
echoHonestly unless you are working with something on a larger scale I don't know if I would recommend using docker here, even the docs advise against it because there is a much smoother automated method for doing things, so just be aware of that when going into things.
Finally, to automate the deployment process, we can do a simple SSH into the VM and then just build and run the containers.
While there should be some premade actions for this, I think it's just nice to see how to do it manually. So you simply add the SSH key to the appropriate file, give it the correct permissions, and then add the hostname (public IP of the VM) to the list of known hosts.
This is where you can get a bit spicy using something like Terraform to define your infrastructure and then running it in case your VM is not active, or you can use whatever CLI tools to define the infrastructure that way. I just added a basic running check.
For the actual deployment, since we set the SSH key, we can simply do ssh username@hostname 'echo "you are in the vm"' to run the required commands in the VM.
These are most of the resources that I used in the processes of learning about how to set this up and getting everything working smoothly