Is there something like a backlog?

[pelcasandra]

I see Unicorn has a handy feature listen ENV['PORT'], :backlog => Integer(ENV['UNICORN_BACKLOG'] || 200) to reduce the time in attempt to try another worker.

Is there something similar on Puma? I'm seeing this as being a mayor problem as more workers / instances are up and running.

[inspire22]

How do I set this? It doesn't seem to work in puma.rb config..

[mscoutermarsh]

@inspire22 I was able to set it by passing a query param to bind. Here's an example config that I tested on Heroku.

workers Integer(ENV['PUMA_WORKERS'] || 3)
threads Integer(ENV['MIN_THREADS']  || 1), Integer(ENV['MAX_THREADS'] || 16)

preload_app!

rackup      DefaultRackup

port = Integer(ENV['PORT'] || 3000)
backlog = Integer(ENV['PUMA_BACKLOG'] || 20)

bind "tcp://0.0.0.0:#{port}?backlog=#{backlog}"

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

on_worker_boot do
  # worker specific setup
  ActiveSupport.on_load(:active_record) do
    config = ActiveRecord::Base.configurations[Rails.env] ||
                Rails.application.config.database_configuration[Rails.env]
    config['pool'] = ENV['MAX_THREADS'] || 16
    ActiveRecord::Base.establish_connection(config)
  end
end

[david-harkness]

A way to do this with nginx would be great