Puma not creating socket at defined location when started with `rails server`

Question

I'm deploying my Rails app using nginx, puma, and capistrano. It's deployed by a user called deploy and the deploy location is under the home directory (/home/deploy)

I have Puma configured to create a socket under the shared folder that Capistrano symlinks all it's releases to. Correspondingly, nginx is configured to look at that socket as well (see config files below)

However when I start up the Rails / Puma webserver -

cd /home/deploy/my_app/current
SECRET_KEY_BASE=.... DATABASE_PASSWORD=... rails s -e production

I notice that no socket file is created. When I visit the site in my browser and then look at the Nginx error log, it is also complaining about that socket not existing.

2016/07/17 14:26:19 [crit] 26055#26055: *12 connect() to unix:/home/deploy/my_app/shared/tmp/sockets/puma.sock failed (2: No such file or directory) while connecting to upstream, client: XX.YY.XX.YY, server: localhost, request: "GET http://testp4.pospr.waw.pl/testproxy.php HTTP/1.1", upstream: "http://unix:/home/deploy/my_app/shared/tmp/sockets/puma.sock:/500.html", host: "testp4.pospr.waw.pl"

How do I go about getting puma to create that socket?

Thanks!

Puma Config

# config/puma.rb

...

# `shared_dir` is the symlinked `shared/` directory created
# by Capistrano - `/home/deploy/my_app/shared`

# Set up socket location
bind "unix://#{shared_dir}/tmp/sockets/puma.sock"

# Logging
stdout_redirect "#{shared_dir}/log/puma.stdout.log", "#{shared_dir}/log/puma.stderr.log", true

# Set master PID and state locations
pidfile "#{shared_dir}/tmp/pids/puma.pid"
state_path "#{shared_dir}/tmp/pids/puma.state"
activate_control_app

...

Nginx sites config

# /etc/nginx/sites-available/default

upstream app {
    # Path to Puma SOCK file
    server unix:/home/deploy/my_app/shared/tmp/sockets/puma.sock fail_timeout=0;
}

server {
    listen 80;
    server_name localhost;

    root /home/deploy/my_app/public;

    try_files $uri/index.html $uri @app;

    location @app {
        proxy_pass http://app;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header Host $http_host;
        proxy_redirect off;
    }

    error_page 500 502 503 504 /500.html;
    client_max_body_size 4G;
    keepalive_timeout 10;
}

Solution

Are you sure you are running Puma with that configuration? I don't think rails server is the proper way to start Puma in a production environment.

I would use this instead:

RACK_ENV=production bundle exec puma -C config/puma.rb

Once you get this working manually, then use the --daemon flag to keep the server running in the background.

Also, where is shared_dir defined in your config/puma.rb? Perhaps you omitted the part of the file, but if not, make sure you insert the correct value.

Comments

[user2490003]

You were absolutely right. I needed to start it with the puma command directly. For some reason I assumed requiring the puma gem would make rails default to using puma as the main server, and so rails server would call puma. Rookie mistake I guess - thanks!