Home / server / Questions / 827381
In Process
Naftuli Kay
Asked: 2017-01-20 11:54:14 +0800 CST

Upstream resolve breaks configuration

  • 772

I am on Ubuntu 16.04 with the latest NGINX installed from their official repository:

$ sudo nginx -v
nginx version: nginx/1.11.8

I have the following defined in /etc/nginx/sites-available/greendot.conf:

# local dnsmasq
resolver 127.0.0.1;

map $http_upgrade $connection_upgrade {
  default upgrade;
  '' close;
}

upstream websocket {
  zone elixr 64k;
  server greendot-elixr-1:4000 resolve;
  server greendot-elixr-2:4000 resolve;
}

server {
  listen 4000;
  location / {
    proxy_pass http://websocket;
    proxy_http_version 1.1;
    proxy_set_header Upgrade $http_upgrade;
    proxy_set_header Connection $connection_upgrade;
  }
}

I have followed the documentation on setting up DNS resolved upstream servers, and yet this configuration fails:

$ sudo nginx -T
nginx: [emerg] invalid parameter "resolve" in /etc/nginx/sites-enabled/greendot-nginx.conf:11
nginx: configuration file /etc/nginx/nginx.conf test failed

My goal is to simply create a server which uses DNS to periodically resolve its members. What am I doing wrong?

nginx

1 Answers

  1. The resolve parameter is part of nginx's commercial offering.

    Here's the approach I ended up using:

    server {
    ...

    # Use docker's internal nameserver
    resolver 127.0.0.11 valid=10s ipv6=off;

    location / {
        set $target myproject-web;
        proxy_pass  http://$target:8000;
    }
}

    Note that:

    • You have to define a variable (like $target), or nginx will query once, and use the resulting IP(s) forever (ignoring the resolver's 'valid' parameter).
    • If the DNS query returns several IPs; nginx will round-robin to them.
    • This approach may not work for you if you have more complex upstream requirements (backup endpoints, etc).
    • 0