How to dockerize #rails app with #puma. Edit config/application.rb and config/puma.rb #docker #ruby

application.rb

module YourApplicationName
  class Application < Rails::Application

    # ...
    
    # We want to set up a custom logger which logs to STDOUT.
    # Docker expects your application to log to STDOUT/STDERR and to be ran
    # in the foreground.
    config.log_level = ENV.fetch('LOG_LEVEL', :debug)
    config.log_tags  = [:subdomain, :uuid]
    config.logger    = ActiveSupport::TaggedLogging.new(Logger.new(STDOUT))

    # ...
  end
end

Procfile

api: bundle exec puma -C config/puma.rb

puma.rb

# Before configuring Puma, you should look up the number of CPU cores your server has, change this to match your CPU core count
workers Integer(ENV['WEB_CONCURRENCY'] || [1, `grep -c processor /proc/cpuinfo`.to_i].max)
threads_count = Integer(ENV['MAX_THREADS'] || 5)
threads threads_count, threads_count

preload_app!

rackup DefaultRackup

# HTTP interface
port 3000

# HTTPS inteface
# How to generate certificate: https://gist.github.com/tadast/9932075
ssl_bind '0.0.0.0', '3001', { key: 'ssl/server.key', cert: 'ssl/server.crt' }

environment ENV['RACK_ENV'] || 'development'

stdout_redirect(stdout = '/dev/stdout', stderr = '/dev/stderr', append = true)

on_worker_boot do
  # Worker specific setup for Rails 4.1+
  # See: https://devcenter.heroku.com/articles/deploying-rails-applications-with-the-puma-web-server#on-worker-boot
  ActiveRecord::Base.establish_connection
end

Any thoughts on Delayed::Jobs?

Thing that within docker such things as Sidekiq, Que or delayed jobs should be running within another container as a separate process, that's how docker suggest to manage things (one process in container), so

FROM andrius/alpine-ruby

# all the stuff you want to add

CMD ["bundle", "exec", "some_delayed_job_application"]

Hi @andrius. Could help to share Dockerfile to build image? And i'm not sure multiple workers can run in single docker process. Help to explain too. Thanks in advance

Many months ago :) Perhaps answer is not relevant, anyways:

docker want us to run a single process, with ruby I do not see any problem, just run two instances in this case, i.e.:

Shared code with all the models, classes and controllers in Dockerfile.base:

FROM ruby:BLAH

WORKDIR /app

COPY . .

RUN bundle install

GUI Dockerfile.gui:

FROM base
CMR ["bundle", "exec", "rails", "server", "0.0.0.0"]

Workers Dockerfile.workers:

FROM base
CMR ["bundle", "exec", "sidekiq", "... args ..."]

docker build --pull --force-rm -t base --file ./Dockerfile.base .
docker build --pull --force-rm -t gui --file ./Dockerfile.gui .
docker build --pull --force-rm -t workers --file ./Dockerfile.workers .

So then we can run them separately. There is more options, i.e. we can use supervisord to handle multiple processes with docker, i.e. based on this example.