Skip to content

karmi/chef-hello-cloud

Repository files navigation

Chef Server Hello Cloud

This repository contains a tutorial for Chef Server.

It deploys a full-stack Ruby On Rails application on the following infrastructure:

  • 1 load balancer running HAproxy
  • 3 application servers
  • 1 database server with PostgreSQL and Redis
  • 2 elasticsearch servers in cluster

Exactly as prescribed:

Hello World in Cloud

The application cookbook installs Ruby 1.9.3 via RVM, tweaks the PostgreSQL settings, searches Chef Server for database and elasticsearch nodes and stores their IPs in the configuration files, restores data from SQL dump and imports it into the search index.

The monitoring cookbook ensures all parts of the stack are monitored via Monit.

There are two options how to install the stack: locally, with Vagrant, and on Amazon EC2.

The application being deployed is Rubygems.org, in a specific version which adds full-text search using elasticsearch. See the rubygems/rubygems.org#455 issue for more information

Prerequisites

First, you need to have a valid Chef Server account, your user validation key, and your organization validation key.

Arguably the easiest way is to create a free account in Hosted Chef, provided by Opscode.

You need to export the following environment variables for the provisioning scripts:

export CHEF_ORGANIZATION='<your organization name>'
export CHEF_ORGANIZATION_KEY='/path/to/your/your-organization-validator.pem'

Building the Stack in Vagrant

Once you have your Chef Server credentials at hand, clone the repo:

git clone git://github.com/karmi/chef-hello-cloud.git
cd chef-hello-cloud

... install the required rubygems:

bundle install

... install the site cookbooks with Berkshelf:

berks install --path ./site-cookbooks/

... upload the cookbooks to Chef Server:

knife cookbook upload --all

... and finally, upload all roles to Chef Server:

knife role from file roles/*.rb

If everything goes fine, build the simplified infrastructure in Vagrant:

time vagrant up

The process should take between 30 and 50 minutes on a reasonable workstation with at least 4GB of RAM. It has been successfuly tested on Mac Book Air Mid 2011, Mac Book Pro Mid 2010 and iMac Mid 2009.

Building the Stack in Amazon EC2

To build the stack in Amazon EC2, in addition to Chef Server credentials, you need to export the path to your private SSH key, downloaded from the AWS console:

export SSH_IDENTITY_FILE='/path/to/your/name-ec2.pem'

You also need to create the following security groups:

  • database with ports 80 and 2812 opened to the outside world
  • elasticsearch with ports 9200 and 9300 opened to the application and elasticsearch groups
  • application with ports 80 and 2812 opened to the outside world
  • load_balancer with ports 80, 2812 and 22002 opened to the outside world

After that, just create the servers with Rake commands.

First, create the database and elasticsearch servers:

time rake server:create NAME=ec2-database ROLE=database
time rake server:create NAME=ec2-elasticsearch-1 ROLE=elasticsearch
time rake server:create NAME=ec2-elasticsearch-2 ROLE=elasticsearch

Once ready, create application servers:

time rake server:create NAME=ec2-application-1 ROLE=application
time rake server:create NAME=ec2-application-2 ROLE=application
time rake server:create NAME=ec2-application-3 ROLE=application

Finally, create the load balancer server:

time rake server:create NAME=ec2-load_balancer ROLE=load_balancer

After the stack is built, open the Public DNS Name URL in your web browser. You should see Rubygems.org with some nice fulltext search features running (hint: click "Tips" on the search page).


(c) 2012 Karel Minarik & Vojtech Hyza. MIT License.

About

A demo of a full stack Rails application deployment (1x load balancer, 3x appserver, 1x database, 3x elasticsearch) with Chef Server

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages