Project

cloudster

0.04
No commit activity in last 3 years
No release in over 3 years
Cloudster is a Ruby gem that was born to cut the learning curve involved in writing your own CloudFormation templates. If you don't know what a CloudFormation template is, but know about the AWS Cloud offerings, you can still use cloudster to provision your stack. Still in infancy , cloudster can create a very basic stack like a breeze. All kinds of contribution welcome !
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
 Dependencies

Development

~> 1.8.8
>= 0

Runtime

= 1.15.0
 Project Readme

Cloudster

The fastest way to provision your AWS stack !

Gem Version Build Status Coverage Status Code Climate Dependency Status still maintained

Cloudster is a Ruby gem that was born to cut the learning curve involved in writing your own CloudFormation templates. If you don't know what a CloudFormation template is, but know about the AWS Cloud, you can still use cloudster to provision your stack like a boss. Checkout the Usage section.

Installation

gem install cloudster

Usage

Create AWS resources :

app_server = Cloudster::Ec2.new(:name => 'AppServer',
  :key_name => 'mykey',
  :image_id => 'ami_image_id',
  :instance_type => 't1.micro',
  :security_groups => ["DevSecurityGroup"]
)

#Well I wanted to add chef clients to my app servers , so ..
chef_client = Cloudster::ChefClient.new(
 :validation_key => 'asd3e33880889098asdnmnnasd8900890a8sdmasdjna9s880808asdnmnasd90-a',
 :server_url => 'http://10.50.60.70:4000',
 :node_name => 'project.environment.appserver_1',
 :interval => 1800
)
chef_client.add_to(app_server)

elastic_ip = Cloudster::ElasticIp.new(:name => 'AppServerElasticIp')
elastic_ip.add_to(app_server)

app_server_2 = Cloudster::Ec2.new(:name => 'AppServer2',
  :key_name => 'mykey',
  :image_id => 'ami_image_id'
)

#Add your app servers to the ElasticLoadBalancer
load_balancer = Cloudster::Elb.new(:name => 'LoadBalancer',
  :instance_names => ['AppServer', 'AppServer2'],
  :listeners => [{:port => 80, :instance_port => 8080, :protocol => 'HTTP'}]
)

database = Cloudster::Rds.new(
  :name => 'MySqlDB',
  :instance_class => 'db.t1.micro',
  :storage_class => '100',
  :username => 'admin',
  :password => 'admin123',
  :engine => 'MySQL',
  :multi_az => true
)

storage = Cloudster::S3.new(
  :name => 'MyBucket'
)

cloud_front = Cloudster::CloudFront.new(:name => 'CloudFrontResource')
cloud_front.add_to storage

elasticache = Cloudster::ElastiCache.new(
  :name => 'CacheResource',
  :node_type => 'cache.t1.micro',
  :cache_security_group_names => ['default'],
  :engine => 'memcached',
  :node_count => 3
)

Create an instance for your cloud :

cloud = Cloudster::Cloud.new(:access_key_id => 'accesskeyid', :secret_access_key => 'topsecretaccesskey', :region => 'us-west-1')

Provision the stack :

cloud.provision(:resources => [app_server, app_server_2, load_balancer, database], :stack_name => 'TestStack', :description => 'Description of the stack')

Get the CloudFormation template for the stack :

cloud.template(:resources => [app_server, app_server_2, load_balancer, database, storage, elasticache], :description => 'Description of the stack')

Get the CloudFormation template for a resource as a Ruby Hash :

app_server.template

Cloudster can also interact with the provisioned AWS Cloud :

  • Update the stack :
cloud.update(:resources => [app_server, app_server_2], :stack_name => 'TestStack', :description => 'Description of the stack')
  • Delete the stack and it's attached resources :
cloud.delete(:stack_name => 'TestStack')
  • Get the output attributes of each resource in the stack :
cloud.outputs(:stack_name => 'TestStack')
  • Describe the events of a stack :
cloud.events(:stack_name => 'TestStack')
  • Describe the attributes of a stack :
cloud.describe(:stack_name => 'TestStack')
  • Describe all resources of a stack :
cloud.resources(:stack_name => 'TestStack')
  • Get the status of a stack :
cloud.status(:stack_name => 'TestStack')
  • Describe the RDS endpoints in a stack :
cloud.get_database_endpoints(:stack_name => 'TestStack')
  • Get the details of all EC2 intances in a stack :
cloud.get_ec2_details(:stack_name => 'TestStack')
  • Get the details of all RDS intances in a stack :
cloud.get_rds_details(:stack_name => 'TestStack')
  • Get the details of all ELB intances in a stack :
cloud.get_elb_details(:stack_name => 'TestStack')
  • Get details of all keypairs created in the AWS account :
cloud.get_key_pairs
  • Get details of all Security Groups created in the AWS account :
cloud.get_security_groups

More coming soon ..

I'm trying to add every AWS resource to cloudster, one by one. If you don't find what you need, let me know and I'll try to get the feature included ASAP, or you can submit a pull request with the feature - that would be awesome!


Contribute

Got some love for Cloudster? Sweet!

Found a bug?

Log the bug in the issue tracker. Be sure to include all relevant information, like the versions of Cloudster and Ruby you're using. A gist of the code that caused the issue as well as any error messages are also very helpful.

Need help?

You can use the Issues page to ask a new question for now. This is how you do it:

  1. Click on New Issue.
  2. Type in your question and submit.

Have a patch?

Bugs and feature requests that include patches are much more likely to get attention. Here are some guidelines that will help ensure your patch can be applied as quickly as possible:

  1. Use Git and GitHub: The easiest way to get setup is to fork the cloudster repo.

  2. Write unit tests: If you add or modify functionality, it must include unit tests. I use RSpec to test cloudster. If you are not an RSpec expert, if you let me know, I can help you write the specs.

  3. Update the README: If the patch adds or modifies a major feature, modify the README.md file to reflect that. Again if you're not an expert with Markdown syntax, it's really easy to learn. Check out Prose.io to try it out.

  4. Push it: Once you're ready, push your changes to a topic branch and add a note to the ticket with the URL to your branch. Or, say something like, "you can find the patch on johndoe/foobranch". I also gladly accept Github pull requests.

NOTE: I will take in whatever I can get. If you prefer to attach diffs in comments on issues, that's fine; but do know that someone will need to take the diff through the process described above and this can hold things up considerably.

License

MIT

Free Software, Forever . YEAH !

Thanks

To Fog for making AWS APIs simple
To Sinatra README for having a nice 'Contribute' section which I'm using(with minor changes) for Cloudster.