I try to get a lot of mileage out of the single AWS t2.micro instance I keep running, which means I have many different projects running on different paths on a single server. Everytime I finish a project, I have to relearn how to add new “locations” (paths) to the NGINX virtual host.

I was really frustrated when I couldn’t figure out how to add a static location for the Jekyll website I created for the game I’m developing. I kept trying to do something like this:

location / {
  root /home/ubuntu/personal-website;
}
location /array {
  root /home/ubuntu/array-website/_site;
}

But everytime I tried to visit [http://ptsteadman.com/array], I would get a 404 error. I tried a bunch of things: I changed the “root” of the the / location to my game website, and it worked. But no matter what I did, after using sudo service nginx restart, trying to visit the /array location still resulted in a 404. I couldn’t add the new route/location.

Eventually, I realized that the text after the slash in the location is the directory that NGINX will try to find in the “root” location. So, location /array will look for the directory (or file) array in whatever directory “root” is set to. So, I created a symlink to the root of my static jekyll site with the command ln -s /home/ubuntu/array-website/_site /home/ubuntu/array, and then I could set up my nginx config file in sites-enabled as below:

location / {
  root /home/ubuntu/personal-website;
}
location /array {
  root /home/ubuntu;
}

I think this stuff is all pretty obvious to someone who really understands NGINX and file serving, but I only touch NGINX when I’ve finished a project and feel impatient to deploy. I think it’s very confusing that NGINX seems to handle paths differently between / locations and and /foo locations, but a real understanding of NGINX might clear things up.