Sinatra Synchrony with Redis connection pooling - ruby

Question

Is this the correct way of handling Redis connection pooling with Sinatra Synchrony?
My gemfile looks like this:
gem 'sinatra-synchrony'
gem 'hiredis'
gem 'redis'
The sinatra server files use the classic style approach, and generally look like so:
require 'sinatra'
require 'sinatra/synchrony'
require 'redis/connection/hiredis'
require 'redis/connection/synchrony'
require 'redis'
redis = EventMachine::Synchrony::ConnectionPool.new(size: 5) do
Redis.new(path: '/tmp/redis.sock')
end
get / do
# lots of redis reads and writes
end
I then launch multiple instances of the same server application, each under a different port, and use nginx to load balance between them.
Is this the proper solution for connection pooling Redis with Sinatra servers?

Answer 1

Here is a working example, I removed sinatra/sinatra because I don't feel it is needed and I couldn't make it works:
Gemfile:
source :rubygems
gem 'thin'
gem 'rack-fiber_pool'
gem 'hiredis'
gem 'sinatra'
gem 'em-synchrony'
gem 'redis'
config.ru:
require 'rubygems'
require 'bundler/setup'
require 'sinatra/base'
require 'redis/connection/synchrony'
require 'redis'
require 'rack/fiber_pool'
class App < Sinatra::Base
set :template_path, '/tmp'
def initialize
super
#redis = EventMachine::Synchrony::ConnectionPool.new(size: 2) do
Redis.new
end
end
get '/' do
#redis.multi do |r|
r.set('v', "value2")
r.set('v2', '43')
end
#redis.get('v')
end
end
use Rack::FiberPool
use Rack::CommonLogger
run App
And run it with (in the same folder):
bundle
bundle exec thin start
In a real application you would remove the application code from the config.ru file and add a require but at least it gives you a start :)