Multiple processes dns resolution

Questions / Help
1

I have a configuration of the main process being an nginx server, and two other processes that are exposed on ports 3000 and 3001. If I ssh in to the two sub processes, I can curl and get from their endpoints. If I try curling by using the internal dns, it doesn’t work (when ssh’ing and through my nginx server I just get bad gateway).

My fly.toml has:

[processes]
    nginx = "nginx -c /app/nginx.conf -g \"daemon off;\""
    web = "npm run deploy"
    rust_htmx = "./rust_htmx"

[[services]]
  protocol = "tcp"
  internal_port = 80 
  processes = ["nginx"]

In case it might help, my nginx config is

http {
    server {
        listen 80;
        listen [::]:80;

        location /htmx/rust {
            proxy_pass http://rust_htmx.process.<app>.internal:3001;
            proxy_set_header Host $host;
        }
 
        location / {
            proxy_pass http://web.process.<app>.internal:3000;
            proxy_set_header Host $host;
        }
    }
}

Here is curling and getting some response but I can’t through the internal dns, only localhost for some reason. This is on the “web” process vm. (I replaced my machineid and app name for privacy/security, not sure if someone can do something with them).

root@<machineid>:/app# curl <app>.internal
<html>
<head><title>502 Bad Gateway</title></head>
<body>
<center><h1>502 Bad Gateway</h1></center>
<hr><center>nginx/1.18.0</center>
</body>
</html>


root@<machineid>:/app# curl <app>.internal:3000
curl: (7) Failed to connect to <app>.internal port 3000: Connection refused


root@<machineid>:/app# curl web.process.<app>.internal:3000
curl: (7) Failed to connect to web.process.<app>.internal port 3000: Connection refused


root@<machineid>:/app# curl localhost:3000
<!DOCTYPE html>
<html lang="en">
    <head>

I just want this nginx server to work as a reverse proxy for my two processes, any help is appreciated!

2

Hey, you need to set the resolver directive on NGINX to fly’s dns server (fdaa::3):

By the way, if I’m not mistaken you should set the app url as a variable to force nginx re-resolve the DNS periodically - otherwise nginx wont find new/updated machines (at least for the free version)

3

Hey thank you for your input!

It seems like it is properly resolving the apps, but they just refuse any connections via .process..internal for some reason even with curl. I changed my fly.toml to expose some of their ports (which is not what I want to end up doing, I want them private), and I was able to hit the endpoints via their exposed public ports- but still when I curl, it says connection refused (rather than could not resolve host)

If I add

resolver fdaa::3;

To my nginx config, it says invalid port for “fdaa::3”. I haven’t worked with nginx before or networking, so I think I’m adding it wrong (I added it under server).

4

It looks like it was just my bad I wasn’t exposing those processes ports on ipv4 and ipv6, I exposed them on ipv6 and it works now! Thank you for your help

1 Like
5

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.