Scoop

a production-ready Sinatra boilerplate project using Corneal, ActiveRecord, Capistrano, Puma & Nginx

Sinatra is a Ruby framework that helps you quickly create web applications, APIs and microservices. Its minimalism not only attracted a lot of developers, but it also inspired Flask in Python and Express in Node.js

If you are developing microservices or web APIs, Sinatra is a great choice. However, bootstraping a Sinatra project that is optimized for development speed as well as is easy to deploy has become non trivial because

Documentation is not easy to find (you mostly find Ruby on Rails specific tips)
You need to write code to set up and connect DB, manage migrations etc.
You need to create a right directory structure yourself and configure it accordingly
You also miss a boot file that you can use with console and with custom rake tasks/scripts
Setting up projects for deployment is tedious and error prone, with a lot of details to be filled in for Capistrano, Puma and Nginx.

Scoop solves it by providing a good boilerplate that can be a starting point for your next project. It solves all the above mentioned problems and comes up with easy to edit configuration files that can help you correctly configure servers.

Scoop uses

Corneal to make scaffolding models, controllers and views
ActiveRecord as database ORM
Capistrano for deployment
Puma as app server
Nginx as a proxy server

In addition,

it is also JSON API ready with JSON, CORS and JSONP support already enabled,
has a RoR like console, and
comes up with example script and rake task that can be used to perform tasks that load the environment

Try Sinatra & Scoop by forking Scoop Repository

Installation & set up

It is very easy to start using scoop as a base for your project, you need to do the following on your development machine

git clone git@github.com:abhinavs/scoop.git APP_NAME (replace APP_NAME with the name of your application, and replace abhinavs with your username if you have forked the repo)
rvm install 3.0.1
gem install corneal
gem install foreman
gem install bundler
cd scoop; bundle install

You should also see config/database.yml file, and change name of the database. After that create the database bundle exec rake db:create

Development

Scoop used Corneal - it is a Sinatra app generator with useful utilities

Generate a model: corneal model NAME
Generate a controller: corneal controller NAME
Generate a controller with no views: corneal controller NAME --no-views
Generate a model with its associated views and controllers: corneal scaffold NAME

Scoop has JSON integration, and it is perfect for APIs. You can check out pincodr's controller to see how to create APIs.

You can migrate data using the following command bundle exec rake db:migrate

Starting development server

You can start the development server using foreman foreman start

If you open the http://localhost:9393 you should see a page getting rendered. You can also see http://localhost:9393/status for JSON response. Both of these routes are inside app/controllers/example_controller.rb

Console & Rake tasks

From the application directory, do bundle exec rake console to open rails like console.

If you want to run create and run custom rake tasks, look at Rakefile and bin/example_script.rb for example

Deploying to production

Our production set up uses

rvm as ruby version manager
We add puma as a sysctl service for easy management

Please note - if you need help in installing RVM and other essential packages, you can check out this guide on Soopr. If you need help in installing database, you can see "Creating A Database" section on Go Rails' guide

Assuming you have created the database, and have installed rvm, install ruby on the server(s) byrvm install 3.0.1

rvm use --default 3.0.1 # You can also set it as default
gem install bundler # Install bundler
sudo apt-get install nginx

To start with first see config/deploy.rb and config/deploy/production.rb files, and change the following details

Server IPs (instances in config/deploy/production.rb)
Name of the application (application)
Git repo URL (repo_url)
Full path of the directory in which you want code to be deployed (deploy_to)
Name of the deploying user (deploy_user) - this user will ssh into the system
SSH key (deploy_ssh_keys)

Installing puma as a service

Modify config/scoop-puma.example.service

Update WorkingDirectory, ExecStart, ExecStop with right working directory
Update SyslogIdentifier with the name you want to give. I generally prefer to keep it as the name of the app and puma (scoop-puma) so that you can deploy multiple services.

Modify config/deploy.rb and replace scoop-puma with the name you have chosen.

Create the directory you have mentioned in your deploy_to variable
Create an .prod-env in that directory (look at config/prod-env) file as an example. You need DATABASE_URL, SINATRA_ENV & RACK_ENV variables. Add any other credentials you have to use.

From your local machine, deploy code by cap production deploy - this step may not succeed, but ensure your code reaches the server.

On the server(s), now we need to install puma as a service (I am copying it as scoop-puma.service, you should use the service name you have given in SyslogIdentifier)

cp config/scoop-puma.example.service /usr/lib/systemd/system/scoop-puma.service
sudo systemctl daemon-reload
sudo systemctl enable scoop-puma.service

From your local machine, deploy code once again using cap production deploy - ideally this time, deployment should succeed and puma server should come up.

Check if puma server is up or not by sudo service scoop-puma status

Config Nginx

Modify `scoop.nginx.conf' and change the following things

Name of the server server_name
Path wherever you see - this path is the server path, and will have current in it.

On the server(s) copy this config file to /etc/nginx/sites-enabled/scoop.conf and restart the nginx (sudo service nginx restart)

(You might have to symlink scoop.conf file in sites-available directory too - you can do that by sudo ln -s /etc/nginx/sites-enabled/scoop.conf /etc/nginx/sites-available/scoop.conf)

Final Deploy

Deploy it once again cap production deploy

contact: abhinav | homepage

Other Projects

If you like Scoop, do check out my other projects

cookie - a free landing website boilerplate using Jekyll and Tailwind CSS
moonwalk - a fast and minimal blog theme with clean dark mode
soopr - a tool that supports you in content marketing
apicagent - a FREE API that extracts device details from user-agent string
pincodr - a FREE API for Indian pincodes
humangous - create public and private 'working with you' guides
blockr - a CLI tool to help you easily block and unblock websites
microrequests - a Python library to help you consume microservice efficiently

✨⚡You can read more about me on my blog or follow me on Twitter- @abhinav

abhinavs/scoop