Rails 5 Action Cable deployment with Nginx, Puma & Redis

Question

I am trying to deploy an Action Cable -enabled-application to a VPS using Capistrano. I am using Puma, Nginx, and Redis (for Cable). After a couple hurdles, I was able to get it working in a local developement environment. I'm using the default in-process /cable URL. But, when I try deploying it to the VPS, I keep getting these two errors in the JS-log:

Establishing connection to host ws://{server-ip}/cable failed.
Connection to host ws://{server-ip}/cable was interrupted while loading the page.

And in my app-specific nginx.error.log I'm getting these messages:

2016/03/10 16:40:34 [info] 14473#0: *22 client 90.27.197.34 closed keepalive connection

Turning on ActionCable.startDebugging() in the JS-prompt shows nothing of interest. Just ConnectionMonitor trying to reopen the connection indefinitely. I'm also getting a load of 301: Moved permanently -requests for /cable in my network monitor.

Things I've tried:

• Using the async adapter instead of Redis. (This is what is used in the developement env)

• Adding something like this to my /etc/nginx/sites-enabled/{app-name}:

  location /cable/ {
    proxy_pass http://puma;
    proxy_http_version 1.1;
    proxy_set_header Upgrade $http_upgrade;
    proxy_set_header Connection "Upgrade";
  }
  

• Setting Rails.application.config.action_cable.allowed_request_origins to the proper host (tried "http://{server-ip}" and "ws://{server-ip}")

• Turning on Rails.application.config.action_cable.disable_request_forgery_protection

No luck. What is causing the issue?

$ rails -v
Rails 5.0.0.beta3

Please inform me of any additional details that may be useful.

Answer 1

Finally, I got it working! I've been trying various things for about a week...

The 301-redirects were caused by nginx actually trying to redirect the browser to /cable/ instead of /cable. This is because I had specified /cable/ instead of /cable in the location stanza! I got the idea from this answer.