nginx load-balancing and reverse-proxy multiports : 502 Bad Gateway

Question

I used two Icecast servers which host many webradios streams. Each stream use a port between 8000 and 9000.

I use nginx (1.16.1 on Debian 10) to :

• easily allow HTTPS for all streams
• automatically change server on failure (ex.: server down). (high-availability / load-balance)

In fact, this is what I wish :

When user listen a stream on https://hosting.mydomain.com:8xxx , I want to transparently (proxify) send the request on one of the two Icecast's server.

Example :

If HTTPS and Icecast_1 is alive then send the request to Icecast_1.

If HTTPS and Icecast_1 is down then send the request to Icecast_2.

To do it, I've set the following :

#Icecast's cluster : upstream backend { ip_hash; keepalive 64; #Icecast 1 : server 10.1.0.101 ; #Icecast 2 : server 10.1.0.102 ; } #SSL for all server { listen 8000-9000 ssl ; server_name hosting.mydomain.com; access_log /var/log/nginx/reverse-ssl-access.log; error_log /var/log/nginx/reverse-ssl-error.log; # ssl on; ssl_protocols TLSv1.2 TLSv1.1 TLSv1 ; ssl_certificate /etc/letsencrypt/live/hosting.mydomain.com/fullchain.pem; ssl_certificate_key /etc/letsencrypt/live/hosting.mydomain.com/privkey.pem; location / { resolver 8.8.8.8; proxy_pass http://backend:$server_port; } } 

I've a problem when I try to do something like this : "http://backend:$server_port".

In the web browser I've the error : 502 Bad gateway

In the errors logs on the server : *1 no resolver defined to resolve backend .

Can you help me ?

Answer 1

You need to set the server port in the upstream configuration. Check if you can use the $server_port in stead of 6789.

upstream backend { ip_hash; keepalive 64; #Icecast 1 : server 10.1.0.101:6789 ; #Icecast 2 : server 10.1.0.102:6789 ; } server { listen 8000-9000 ssl ; server_name nginx.mydomain.com; access_log /var/log/nginx/reverse-ssl-access.log; error_log /var/log/nginx/reverse-ssl-error.log; # ssl on; ssl_protocols TLSv1.2 TLSv1.1 TLSv1 ; ssl_certificate /etc/letsencrypt/live/hosting.mydomain.com/fullchain.pem; ssl_certificate_key /etc/letsencrypt/live/hosting.mydomain.com/privkey.pem; location / { proxy_pass http://backend; } } 

If you want to use $server_port, you can't use upstream, so, no load balancing.

 location / { proxy_pass http://10.1.0.101:$server_port; } 

Answer 2

@Gerard H. Pill , this is my current server's configuration :

#Icecast's cluster : upstream backend { ip_hash; keepalive 64; #Icecast 1 : server 10.1.0.101:8014 ; #Icecast 2 : server 10.1.0.102:8014 ; } #SSL for all server { listen 8000-9000 ssl ; server_name hosting.mydomain.com; access_log /var/log/nginx/reverse-ssl-access.log; error_log /var/log/nginx/reverse-ssl-error.log; # ssl on; ssl_protocols TLSv1.2 TLSv1.1 TLSv1 ; ssl_certificate /etc/letsencrypt/live/hosting.mydomain.com/fullchain.pem; ssl_certificate_key /etc/letsencrypt/live/hosting.mydomain.com/privkey.pem; location / { resolver 8.8.8.8; proxy_pass http://backend; } } 

I saved it and I restarted nginx.

Answer 3

@Gerard H. Pille , Thank you for your answers and your time. You helped me a lot