I have several nginx services configured on my system. Suppose one of these services is configured with upstream block and with server directive pointing to fqdn.

With this configuration files, when nginx starts up or reloads config file and if it's not able to resolve fqdn, then it wouldn't even start or reload rest of the configuration data. As a result, all my services are down just because one service encounters an error.

With set directive and using a variable in proxy_pass directive, you can make nginx to resolve the fqdn only during request proxying. But in my case, I have a upstream block and using server directive to specify fqdn. So that solution will not work for me. So is there a way to make nginx still go ahead with the configuration and start even when it encounters DNS error or make nginx delay DNS resolution only when the http request arrives?


As an example, here is the sample block inside a conf file

upstream ustreamtest{

server test-server.com:80 ;
}

server {
server_name test ;
listen 192.168.150.89:802 ;
resolver 10.0.0.5 ;
proxy_bind 192.168.150.87;

location / {
set $test "ustreamtest";
proxy_pass HTTP://$test;

proxy_next_upstream error timeout invalid_header http_500 http_502 http_503 http_504;

include /usr/local/nginx/conf/proxy.conf ;
proxy_redirect off;
}
}