Unleash::Client

Ruby client for the Unleash feature management service.

Migrating to v6

If you use custom strategies or override built-in ones, read the complete migration guide before upgrading to v6.

Supported Ruby interpreters

• MRI 3.3
• MRI 3.2
• MRI 3.1
• MRI 3.0
• MRI 2.7
• jruby 9.4

Installation

Add this line to your application's Gemfile:

gem 'unleash', '~> 6.0.10'

And then execute:

$ bundle

Or install it yourself as:

$ gem install unleash

Configuration

It is required to configure:

• app_name with the name of the running application
• url of your Unleash server
• custom_http_headers with {'Authorization': '<YOUR_API_TOKEN>'} when using Unleash v4+

It is highly recommended to configure:

• instance_id parameter with a unique identifier for the running instance

Unleash.configure do |config|
  config.app_name            = 'my_ruby_app'
  config.url                 = '<YOUR_UNLEASH_URL>/api'
  config.custom_http_headers = {'Authorization': '<YOUR_API_TOKEN>'}
end

or instantiate the client with the valid configuration:

UNLEASH = Unleash::Client.new(url: '<YOUR_UNLEASH_URL>/api', app_name: 'my_ruby_app', custom_http_headers: {'Authorization': '<YOUR_API_TOKEN>'})

Dynamic custom HTTP headers

If you need custom HTTP headers that change during the lifetime of the client, you can pass custom_http_headers as a Proc.

Unleash.configure do |config|
  config.app_name            = 'my_ruby_app'
  config.url                 = '<YOUR_UNLEASH_URL>/api'
  config.custom_http_headers =  proc do
    {
      'Authorization': '<YOUR_API_TOKEN>',
      'X-Client-Request-Time': Time.now.iso8601
    }
  end
end

List of arguments

Argument	Description	Required?	Type	Default value
url	Unleash server URL.	Y	String	N/A
app_name	Name of your program.	Y	String	N/A
instance_id	Identifier for the running instance of your program—set this to be able trace where metrics are being collected from.	N	String	random UUID
environment	Unleash context option, for example, prod or dev. Not yet in use. Not the same as the SDK's Unleash environment.	N	String	default
project_name	Name of the project to retrieve feature flags from. If not set, all feature flags will be retrieved.	N	String	nil
refresh_interval	How often the Unleash client should check with the server for configuration changes.	N	Integer	15
metrics_interval	How often the Unleash client should send metrics to server.	N	Integer	60
disable_client	Disables all communication with the Unleash server, effectively taking it offline. If set, is_enabled? always answer with the default_value and configuration validation is skipped. Will also forcefully set disable_metrics to true. Defeats the entire purpose of using Unleash, except when running tests.	N	Boolean	false
disable_metrics	Disables sending metrics to Unleash server. If the disable_client option is set to true, then this option will also be set to true, regardless of the value provided.	N	Boolean	false
custom_http_headers	Custom headers to send to Unleash. As of Unleash v4.0.0, the Authorization header is required. For example: {'Authorization': '<YOUR_API_TOKEN>'}.	N	Hash/Proc	{}
timeout	How long to wait for the connection to be established or wait in reading state (open_timeout/read_timeout)	N	Integer	30
retry_limit	How many consecutive failures in connecting to the Unleash server are allowed before giving up. The default is to retry indefinitely.	N	Float::INFINITY	5
backup_file	Filename to store the last known state from the Unleash server. It is best to not change this from the default.	N	String	Dir.tmpdir + "/unleash-#{app_name}-repo.json
logger	Specify a custom Logger class to handle logs for the Unleash client.	N	Class	Logger.new(STDOUT)
log_level	Change the log level for the Logger class. Constant from Logger::Severity.	N	Constant	Logger::WARN
bootstrap_config	Bootstrap config for loading data on startup—useful for loading large states on startup without (or before) hitting the network.	N	Unleash::Bootstrap::Configuration	nil
strategies	Strategies manager that holds all strategies and allows to add custom strategies.	N	Unleash::Strategies	Unleash::Strategies.new

For a more in-depth look, please see lib/unleash/configuration.rb.

Environment Variable	Description
UNLEASH_BOOTSTRAP_FILE	File to read bootstrap data from
UNLEASH_BOOTSTRAP_URL	URL to read bootstrap data from

Usage in a plain Ruby application

require 'unleash'
require 'unleash/context'

@unleash = Unleash::Client.new(app_name: 'my_ruby_app', url: '<YOUR_UNLEASH_URL>/api', custom_http_headers: { 'Authorization': '<YOUR_API_TOKEN>' })

feature_name = "AwesomeFeature"
unleash_context = Unleash::Context.new
unleash_context.user_id = 123

if @unleash.is_enabled?(feature_name, unleash_context)
  puts " #{feature_name} is enabled according to unleash"
else
  puts " #{feature_name} is disabled according to unleash"
end

if @unleash.is_disabled?(feature_name, unleash_context)
  puts " #{feature_name} is disabled according to unleash"
else
  puts " #{feature_name} is enabled according to unleash"
end

Usage in a Rails application

1. Add Initializer

The initializer setup varies depending on whether you’re using a standard setup, Puma in clustered mode, Phusion Passenger, or Sidekiq.

1.a Initializer for standard Rails applications

Put in config/initializers/unleash.rb:

Unleash.configure do |config|
  config.app_name = Rails.application.class.module_parent_name
  config.url      = '<YOUR_UNLEASH_URL>'
  # config.instance_id = "#{Socket.gethostname}"
  config.logger   = Rails.logger
  config.custom_http_headers = {'Authorization': '<YOUR_API_TOKEN>'}
end

UNLEASH = Unleash::Client.new

# Or if preferred:
# Rails.configuration.unleash = Unleash::Client.new

For config.instance_id use a string with a unique identification for the running instance. For example, it could be the hostname if you only run one App per host, or the docker container ID, if you are running in Docker. If not set, the client will generate a unique UUID for each execution.

To have it available in the rails console command as well, also add to the file above:

Rails.application.console do
  UNLEASH = Unleash::Client.new
  # or
  # Rails.configuration.unleash = Unleash::Client.new
end

1.b Add Initializer if using Puma in clustered mode

That is, multiple workers configured in puma.rb:

workers ENV.fetch("WEB_CONCURRENCY") { 2 }

with preload_app!

Then you may keep the client configuration still in config/initializers/unleash.rb:

Unleash.configure do |config|
  config.app_name    = Rails.application.class.parent.to_s
  config.url                 = '<YOUR_UNLEASH_URL>/api'
  config.custom_http_headers = {'Authorization': '<YOUR_API_TOKEN>'}
end

But you must ensure that the Unleash client is instantiated only after the process is forked. This is done by creating the client inside the on_worker_boot code block in puma.rb as below:

#...
preload_app!
#...

on_worker_boot do
  # ...

  ::UNLEASH = Unleash::Client.new
end

on_worker_shutdown do
  ::UNLEASH.shutdown
end

without preload_app!

By not using preload_app!:

• The Rails constant will not be available.
• Phased restarts will be possible.

You need to ensure that in puma.rb:

• The Unleash SDK is loaded with require 'unleash' explicitly, as it will not be pre-loaded.
• All parameters are set explicitly in the on_worker_boot block, as config/initializers/unleash.rb is not read.
• There are no references to Rails constant, as that is not yet available.

Example for puma.rb:

require 'unleash'

#...
# no preload_app!

on_worker_boot do
  # ...

  ::UNLEASH = Unleash::Client.new(
    app_name: 'my_rails_app',
    url: '<YOUR_UNLEASH_URL>/api',
    custom_http_headers: {'Authorization': '<YOUR_API_TOKEN>'},
  )
end

on_worker_shutdown do
  ::UNLEASH.shutdown
end

Note that we also added shutdown hooks in on_worker_shutdown, to ensure a clean shutdown.

1.c Add Initializer if using Phusion Passenger

The Unleash client needs to be configured and instantiated inside the PhusionPassenger.on_event(:starting_worker_process) code block due to smart spawning:

The initializer in config/initializers/unleash.rb should look like:

PhusionPassenger.on_event(:starting_worker_process) do |forked|
  if forked
    Unleash.configure do |config|
      config.app_name    = Rails.application.class.parent.to_s
      # config.instance_id = "#{Socket.gethostname}"
      config.logger      = Rails.logger
      config.url                 = '<YOUR_UNLEASH_URL>/api'
      config.custom_http_headers = {'Authorization': '<YOUR_API_TOKEN>'}
    end

    UNLEASH = Unleash::Client.new
  end
end

1.d Add Initializer hooks when using within Sidekiq

Note that in this case, we require that the code block for Unleash.configure is set beforehand. For example in config/initializers/unleash.rb.

Sidekiq.configure_server do |config|
  config.on(:startup) do
    UNLEASH = Unleash::Client.new
  end

  config.on(:shutdown) do
    UNLEASH.shutdown
  end
end

2. Set Unleash::Context

Add the following method and callback in the application controller to have @unleash_context set for all requests:

Add in app/controllers/application_controller.rb:

  before_action :set_unleash_context

  private
  def set_unleash_context
    @unleash_context = Unleash::Context.new(
      session_id: session.id,
      remote_address: request.remote_ip,
      user_id: session[:user_id]
    )
  end

Alternatively, you can add this method only to the controllers that use Unleash.

3. Sample usage

Then wherever in your application that you need a feature toggle, you can use:

if UNLEASH.is_enabled? "AwesomeFeature", @unleash_context
  puts "AwesomeFeature is enabled"
end

or if client is set in Rails.configuration.unleash:

if Rails.configuration.unleash.is_enabled? "AwesomeFeature", @unleash_context
  puts "AwesomeFeature is enabled"
end

If you don't want to check a feature is disabled with unless, you can also use is_disabled?:

# so instead of:
unless UNLEASH.is_enabled? "AwesomeFeature", @unleash_context
  puts "AwesomeFeature is disabled"
end

# it might be more intelligible:
if UNLEASH.is_disabled? "AwesomeFeature", @unleash_context
  puts "AwesomeFeature is disabled"
end

If the feature is not found in the server, it will by default return false. However, you can override that by setting the default return value to true:

if UNLEASH.is_enabled? "AwesomeFeature", @unleash_context, true
  puts "AwesomeFeature is enabled by default"
end
# or
if UNLEASH.is_disabled? "AwesomeFeature", @unleash_context, true
  puts "AwesomeFeature is disabled by default"
end

Another possibility is to send a block, Lambda or Proc to evaluate the default value:

net_check_proc = proc do |feature_name, context|
  context.remote_address.starts_with?("10.0.0.")
end

if UNLEASH.is_enabled?("AwesomeFeature", @unleash_context, &net_check_proc)
  puts "AwesomeFeature is enabled by default if you are in the 10.0.0.* network."
end

or

awesomeness = 10
@unleash_context.properties[:coolness] = 10

if UNLEASH.is_enabled?("AwesomeFeature", @unleash_context) { |feat, ctx| awesomeness >= 6 && ctx.properties[:coolness] >= 8 }
  puts "AwesomeFeature is enabled by default if both the user has a high enough coolness and the application has a high enough awesomeness"
end

Note:

• The block/lambda/proc can use the feature name and context as arguments.
• The client will evaluate the fallback function once per call of is_enabled(). Please keep this in mind when creating your fallback function.
• The returned value of the block should be a boolean. However, the client will coerce the result to a boolean via !!.
• If both a default_value and fallback_function are supplied, the client will define the default value by ORing the default value and the output of the fallback function.

Alternatively by using if_enabled (or if_disabled) you can send a code block to be executed as a parameter:

UNLEASH.if_enabled "AwesomeFeature", @unleash_context, true do
  puts "AwesomeFeature is enabled by default"
end

Note: if_enabled (and if_disabled) only support default_value, but not fallback_function.

Variations

If no flag is found in the server, use the fallback variant.

fallback_variant = Unleash::Variant.new(name: 'default', enabled: true, payload: {"color" => "blue"})
variant = UNLEASH.get_variant "ColorVariants", @unleash_context, fallback_variant

puts "variant color is: #{variant.payload.fetch('color')}"

Bootstrapping

Bootstrap configuration allows the client to be initialized with a predefined set of toggle states. Bootstrapping can be configured by providing a bootstrap configuration when initializing the client.

@unleash = Unleash::Client.new(
    url: '<YOUR_UNLEASH_URL>/api',
    app_name: 'my_ruby_app',
    custom_http_headers: { 'Authorization': '<YOUR_API_TOKEN>' },
    bootstrap_config: Unleash::Bootstrap::Configuration.new({
        url: "<YOUR_UNLEASH_URL>/api/client/features",
        url_headers: {'Authorization': '<YOUR_API_TOKEN>'}
    })
)

The Bootstrap::Configuration initializer takes a hash with one of the following options specified:

• file_path - An absolute or relative path to a file containing a JSON string of the response body from the Unleash server. This can also be set through the UNLEASH_BOOTSTRAP_FILE environment variable.
• url - A url pointing to an Unleash server's features endpoint, the code sample above is illustrative. This can also be set through the UNLEASH_BOOTSTRAP_URL environment variable.
• url_headers - Headers for the GET HTTP request to the url above. Only used if the url parameter is also set. If this option isn't set then the bootstrapper will use the same url headers as the Unleash client.
• data - A raw JSON string as returned by the Unleash server.
• block - A lambda containing custom logic if you need it, an example is provided below.

You should only specify one type of bootstrapping since only one will be invoked and the others will be ignored. The order of preference is as follows:

• Select a data bootstrapper if it exists.
• If no data bootstrapper exists, select the block bootstrapper.
• If no block bootstrapper exists, select the file bootstrapper from either parameters or the specified environment variable.
• If no file bootstrapper exists, then check for a URL bootstrapper from either the parameters or the specified environment variable.

Example usage:

First, save the toggles locally:

curl -H 'Authorization: <YOUR_API_TOKEN>' -XGET '<YOUR_UNLEASH_URL>/api' > ./default-toggles.json

Then use them on startup:

custom_boostrapper = lambda {
  File.read('./default-toggles.json')
}

@unleash = Unleash::Client.new(
    app_name: 'my_ruby_app',
    url: '<YOUR_UNLEASH_URL>/api',
    custom_http_headers: { 'Authorization': '<YOUR_API_TOKEN>' },
    bootstrap_config: Unleash::Bootstrap::Configuration.new({
        block: custom_boostrapper
    })
)

This example could be easily achieved with a file bootstrapper, this is just to illustrate the usage of custom bootstrapping. Be aware that the client initializer will block until bootstrapping is complete.

Client methods

Method name	Description	Return type
is_enabled?	Checks if a feature toggle is enabled or not	Boolean
enabled?	A more idiomatic Ruby alias for the is_enabled? method	Boolean
if_enabled	Runs a code block, if a feature is enabled	yield
is_disabled?	Checks if feature toggle is enabled or not	Boolean
disabled?	A more idiomatic Ruby alias for the is_disabled? method	Boolean
if_disabled	Runs a code block, if a feature is disabled	yield
get_variant	Gets variant for a given feature	Unleash::Variant
shutdown	Saves metrics to disk, flushes metrics to server, and then kills ToggleFetcher and MetricsReporter threads—a safe shutdown, not generally needed in long-running applications, like web applications	nil
shutdown!	Kills ToggleFetcher and MetricsReporter threads immediately	nil

For the full method signatures, see client.rb.

Local test client

# cli unleash client:
bundle exec bin/unleash-client --help

# or a simple sample implementation (with values hardcoded):
bundle exec examples/simple.rb

Available strategies

This client comes with all the required strategies out of the box:

• ApplicationHostnameStrategy
• DefaultStrategy
• FlexibleRolloutStrategy
• GradualRolloutRandomStrategy
• GradualRolloutSessionIdStrategy
• GradualRolloutUserIdStrategy
• RemoteAddressStrategy
• UnknownStrategy
• UserWithIdStrategy

Custom strategies

You can add custom activation strategies using configuration. In order for the strategy to work correctly it should support two methods name and is_enabled?.

class MyCustomStrategy
  def name
    'myCustomStrategy'
  end

  def is_enabled?(params = {}, context = nil)
    true
  end
end

Unleash.configure do |config|
  config.strategies.add(MyCustomStrategy.new)
end

Development

After checking out the repo, run bin/setup to install dependencies. Then, run bundle exec rake spec to run the tests. You can also run bin/console for an interactive prompt that will allow you to experiment.

This SDK is also built against the Unleash Client Specification tests. To run the Ruby SDK against this test suite, you'll need to have a copy on your machine, you can clone the repository directly using:

git clone --branch v$(ruby echo_client_spec_version.rb) https://github.com/Unleash/client-specification.git

After doing this, bundle exec rake spec will also run the client specification tests.

To install this gem onto your local machine, run bundle exec rake install.

Releasing

To release a new version, follow these steps:

1. Update version number:
   • Increment the version number in the ./lib/unleash/version.rb file according to Semantic Versioning guidelines.
2. Update documentation:
   • If the update includes a major or minor version change, update the Installation section in README.md.
   • Update CHANGELOG.md following the format on Keep a Changelog.
3. Commit changes:
   • Commit the changes with a message like: chore: bump version to x.y.z.
4. Release the gem:
   • On the main branch, run bundle exec rake release to create a git tag for the new version, push commits and tags to origin, and publish .gem file to rubygems.org.

Contributing

Bug reports and pull requests are welcome on GitHub at https://github.com/unleash/unleash-client-ruby.

Be sure to run both bundle exec rspec and bundle exec rubocop in your branch before creating a pull request.

Please include tests with any pull requests, to avoid regressions.

Check out our guide for more information on how to build and scale feature flag systems.