Production deploy gives 404 on /admin (not on /api)

[falko100]

Hey guys,

So I've setup payload to run over https://domainx.dev through an nginx proxy.
In my .env (ecosystem.config.js) I've got "SERVER_URL": "https://domainx.dev",.
In nginx I'm doing proxy_pass http://localhost:3000;

When I got to https://domainx.dev/api/pages I get a result, but when I go to https://domainx.dev/admin I get Cannot GET /admin.
This does work without production mode.

Any tips? I'm slowly giving up hope 🤯

[jmikrut]

Hey @falko100 —

Looks like there is a similar discussion here.

Here's what a typical NGINX config looks like when we deploy to production:

server {
  index index.html;

  server_name mywebsite.com;

  location / {
    proxy_pass http://localhost:1337;
    proxy_http_version 1.1;
    proxy_set_header Upgrade $http_upgrade;
    proxy_set_header Connection 'upgrade';
    proxy_set_header Host $host;
    proxy_cache_bypass $http_upgrade;
    proxy_set_header X-Forwarded-For $remote_addr;
  }

  listen 443 ssl; # managed by Certbot
  ssl_certificate /etc/letsencrypt/live/mywebsite.com/fullchain.pem; # managed by Certbot
  ssl_certificate_key /etc/letsencrypt/live/mywebsite.com/privkey.pem; # managed by Certbot
  include /etc/letsencrypt/options-ssl-nginx.conf; # managed by Certbot
  ssl_dhparam /etc/letsencrypt/ssl-dhparams.pem; # managed by Certbot
}

server {
  if ($host = mywebsite.com) {
      return 301 https://$host$request_uri;
  } # managed by Certbot

  listen 80;

  server_name mywebsite.com;
  return 404; # managed by Certbot
}

One thing you could try is naming your SERVER_URL environment variable to PAYLOAD_PUBLIC_SERVER_URL. By default, Payload does not expose ANY environment variables to the admin panel for security reasons. Instead, we only expose variables prefixed with PAYLOAD_PUBLIC_. Here are some docs about this.

I'm not sure, but if your NGINX config looks similar to ours, maybe your SERVER_URL has something to do with your issue.

Last thing I'd check: can you serve in production locally? Like, running build and then serving locally? You may be having a Webpack error, where your admin panel is not successfully building. That would cause an /admin 404. You need to have successfully run yarn build or similar before you can serve in production.

We'll figure it out!

[falko100]

Steps to reproduce:

1. $ npx create-payload-app
2. $ cd prod-test
3. $ yarn build
4. $ cp .env dist && cd dist
5. $ NODE_ENV=production node server.js

Works if I run the server with $ node server.js

[jmikrut]

Ah, figured it out! Thanks for the steps here.

To determine where the build files are, we are currently doing the following:

path.resolve(process.cwd(), 'build')

In your case, you're running the server in production directly from the dist folder rather than from an npm script or similar from the root folder. This also causes you to need to duplicate your .env file to the dist folder when normally you shouldn't need to do that.

I think what we can do is improve the way that we are finding the build folder, rather than relying on process.cwd() like that, which will make this more bulletproof. But in the meantime, you could adjust your pm2 script to execute using the cwd set to the root of the Payload folder. That will solve this!

[falko100]

Ahh ok. I can do that, but then I get an error that the payload.config.js is missing.

[falko100]

Steps to reproduce:

$ npx create-payload-app
$ cd prod-test
$ yarn build
$ NODE_ENV=production node dist/server.js

[falko100]

This can be solved by adding environment variable; PAYLOAD_CONFIG_PATH=dist/payload.config.js
Server works great now. Thanks @jmikrut for the help (and the amazing CMS, love it! ❤️)

[falko100]

Answered by @jmikrut.
A new problem arised, but the initial problem is resolved.

Ah, figured it out! Thanks for the steps here.

To determine where the build files are, we are currently doing the following:

path.resolve(process.cwd(), 'build')

In your case, you're running the server in production directly from the dist folder rather than from an npm script or similar from the root folder. This also causes you to need to duplicate your .env file to the dist folder when normally you shouldn't need to do that.

I think what we can do is improve the way that we are finding the build folder, rather than relying on process.cwd() like that, which will make this more bulletproof. But in the meantime, you could adjust your pm2 script to execute using the cwd set to the root of the Payload folder. That will solve this!