NGINX integration
Unit is a potent and versatile server in its own right. However, if you’re used to NGINX’s rich feature set, you can deploy it in front of Unit; one notable use case for NGINX here is securing the Unit control socket.
Configure a listener in Unit:
{ "127.0.0.1:8080": { "comment_127.0.0.1:8080": "Socket address where NGINX proxies requests", "pass": "...", "comment_pass": "Unit's internal request destination", "forwarded": { "client_ip": "X-Forwarded-For", "comment_client_ip": "The header field set by NGINX", "source": [ "127.0.0.1" ], "comment_source": "The IP address where NGINX runs" } } }
Here, forwarded is optional; it enables identifying the originating IPs of requests proxied from source.
In NGINX configuration, create an upstream in the http context, adding the listener’s socket as a server:
http { upstream unit_backend { server 127.0.0.1:8080; # Unit's listener socket address } server { location /unit/ { # Arbitrary location proxy_pass http://unit_backend; proxy_set_header Host $host; proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; # Unit's listener must list the same name in client_ip/header } } }
A more compact alternative would be a direct proxy_pass in your location:
http { server { location /unit/ { # Arbitrary location proxy_pass http://127.0.0.1:8080; # Unit's listener socket address proxy_set_header Host $host; proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; # Unit's listener must list the same name in client_ip/header } } }
The proxy_set_header X-Forwarded-For directives work together with the listener’s client_ip option.
For details, see the NGINX documentation. Commercial support and advanced features are also available.
By default, Unit exposes its control API via a UNIX domain socket. These sockets aren’t network accessible, so the API is local only. To enable secure remote access, you can use NGINX as a reverse proxy.
Avoid exposing an unprotected control socket to public networks. Use NGINX or a different solution such as SSH for security and authentication.
Use this configuration template for NGINX (replace placeholders in ssl_certificate, ssl_certificate_key, ssl_client_certificate, allow, auth_basic_user_file, and proxy_pass with real values):
server { # Configure SSL encryption listen 443 ssl; ssl_certificate /path/to/ssl/cert.pem; # Path to your PEM file; use a real path in your configuration ssl_certificate_key /path/to/ssl/cert.key; # Path to your key file; use a real path in your configuration # SSL client certificate validation ssl_client_certificate /path/to/ca.pem; # Path to certification authority PEM file; use a real path in your configuration ssl_verify_client on; # Network ACLs allow 1.2.3.4; # Replicate and update as needed with allowed IPs and network CIDRs deny all; # HTTP Basic authentication auth_basic on; auth_basic_user_file /path/to/htpasswd; # Path to your htpasswd file location / { proxy_pass http://unix:/path/to/control.unit.sock; # Path to Unit's control socket } }
The same approach works for an IP-based control socket:
location / { proxy_pass http://127.0.0.1:8080; # Unit's control socket address }