fog is the Ruby cloud services library, top to bottom:

  • Collections provide a simplified interface, making clouds easier to work with and switch between.
  • Requests allow power users to get the most out of the features of each individual cloud.
  • Mocks make testing and integrating a breeze.

Build Status Dependency Status Code Climate Coverage Status Gem Version Gittip

Getting Started

sudo gem install fog

Now type fog to try stuff, confident that fog will let you know what to do. Here is an example of wading through server creation for Amazon Elastic Compute Cloud:

>> server = Compute[:aws].servers.create
ArgumentError: image_id is required for this operation

>> server = Compute[:aws].servers.create(:image_id => 'ami-5ee70037')
<Fog::AWS::EC2::Server [...]>

>> server.destroy # cleanup after yourself or regret it, trust me

Ruby 1.8.7

The maintainers of this project, in concert with the maintainers of Ruby, strongly recommend using the latest patchlevel of Ruby 1.9.2 or later. As of July 1, 2013, Ruby 1.8.7 is no longer officially maintained. This means fixes will no longer be provided, even for known security vulnerabilities.

With this caveat, if you wish to bundle fog into your application on Ruby 1.8.7, you must add the following line to your Gemfile.

gem 'nokogiri', '~>1.5.0'

Also, ensure that you are using LibXML version 2.8.0, since there is an issue with LibXML version 2.9.0 (and 2.9.1).


A high level interface to each cloud is provided through collections, such as images and servers. You can see a list of available collections by calling collections on the connection object. You can try it out using the fog command:

>> Compute[:aws].collections
[:addresses, :directories, ..., :volumes, :zones]

Some collections are available across multiple providers:

  • compute providers have flavors, images and servers
  • dns providers have zones and records
  • storage providers have directories and files

Collections share basic CRUD type operations, such as:

  • all - fetch every object of that type from the provider.
  • create - initialize a new record locally and a remote resource with the provider.
  • get - fetch a single object by it's identity from the provider.
  • new - initialize a new record locally, but do not create a remote resource with the provider.

As an example, we'll try initializing and persisting a Rackspace Cloud server:

require 'fog'

compute =
  :provider           => 'Rackspace',
  :rackspace_api_key  => key,
  :rackspace_username => username

# boot a gentoo server (flavor 1 = 256, image 3 = gentoo 2008.0)
server = compute.servers.create(:flavor_id => 1, :image_id => 3, :name => 'my_server')
server.wait_for { ready? } # give server time to boot


server.destroy # cleanup after yourself or regret it, trust me


Many of the collection methods return individual objects, which also provide common methods:

  • destroy - will destroy the persisted object from the provider
  • save - persist the object to the provider
  • wait_for - takes a block and waits for either the block to return true for the object or for a timeout (defaults to 10 minutes)


As you might imagine, testing code using Fog can be slow and expensive, constantly turning on and and shutting down instances. Mocking allows skipping this overhead by providing an in memory representation resources as you make requests. Enabling mocking easy to use, before you run other commands, simply run:


Then proceed as usual, if you run into unimplemented mocks, fog will raise an error and as always contributions are welcome!


Requests allow you to dive deeper when the models just can't cut it. You can see a list of available requests by calling #requests on the connection object.

For instance, ec2 provides methods related to reserved instances that don't have any models (yet). Here is how you can lookup your reserved instances:

$ fog
>> Compute[:aws].describe_reserved_instances
#<Excon::Response [...]>

It will return an excon response, which has body, headers and status. Both return nice hashes.

Go forth and conquer

Play around and use the console to explore or check out and the provider documentation for more details and examples. Once you are ready to start scripting fog, here is a quick hint on how to make connections without the command line thing to help you.

# create a compute connection
compute = => 'AWS', :aws_access_key_id => ACCESS_KEY_ID, :aws_secret_access_key => SECRET_ACCESS_KEY)
# compute operations go here

# create a storage connection
storage = => 'AWS', :aws_access_key_id => ACCESS_KEY_ID, :aws_secret_access_key => SECRET_ACCESS_KEY)
# storage operations go here

geemus says: "That should give you everything you need to get started, but let me know if there is anything I can do to help!"


Fog library aims to adhere to Semantic Versioning 2.0.0, although it does not address challenges of multi-provider libraries. Semantic versioning is only guaranteed for the common API, not any provider-specific extensions. You may also need to update your configuration from time to time (even between Fog releases) as providers update or deprecate services.

However, we still aim for forwards compatibility within Fog major versions. As a result of this policy, you can (and should) specify a dependency on this gem using the Pessimistic Version Constraint with two digits of precision. For example:

spec.add_dependency 'fog', '~> 1.0'

This means your project is compatible with Fog 1.0 up until 2.0. You can also set a higher minimum version:

spec.add_dependency 'fog', '~> 1.16'

Getting Help


Please refer to


Please refer to