Unleash client so you can roll out your features with confidence.
Leverage the Unleash Server for powerful feature toggling in your ruby/rails applications.
- MRI 3.3
- MRI 3.2
- MRI 3.1
- MRI 3.0
- MRI 2.7
- MRI 2.6
- jruby 9.4
- jruby 9.3
Add this line to your application's Gemfile:
gem 'unleash', '~> 5.0.4'
And then execute:
$ bundle
Or install it yourself as:
$ gem install unleash
It is required to configure:
url
of the unleash serverapp_name
with the name of the runninng application.custom_http_headers
with{'Authorization': '<API token>'}
when using Unleash v4.0.0 and later.
Please substitute the example 'https://unleash.herokuapp.com/api'
for the url of your own instance.
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 = 'https://unleash.herokuapp.com/api'
config.custom_http_headers = {'Authorization': '<API token>'}
end
or instantiate the client with the valid configuration:
UNLEASH = Unleash::Client.new(url: 'https://unleash.herokuapp.com/api', app_name: 'my_ruby_app', custom_http_headers: {'Authorization': '<API token>'})
If you need custom HTTP headers that change during the lifetime of the client, the custom_http_headers
can be given as a Proc
.
Unleash.configure do |config|
config.app_name = 'my_ruby_app'
config.url = 'https://unleash.herokuapp.com/api'
config.custom_http_headers = proc do
{
'Authorization': '<API token>',
'X-Client-Request-Time': Time.now.iso8601
}
end
end
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 program. Important so you can trace back to where metrics are being collected from. Highly recommended be be set. | N | String | random UUID |
environment |
Unleash context option. Could be 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 features 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? will 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': '<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. 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 on how to loaded data on start-up. This is 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 |
require 'unleash'
require 'unleash/context'
@unleash = Unleash::Client.new(app_name: 'my_ruby_app', url: 'https://unleash.herokuapp.com/api', custom_http_headers: { 'Authorization': '<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
Put in config/initializers/unleash.rb
:
Unleash.configure do |config|
config.app_name = Rails.application.class.parent.to_s
config.url = 'https://unleash.herokuapp.com/api'
# config.instance_id = "#{Socket.gethostname}"
config.logger = Rails.logger
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 it is not set the client will generate an 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
Add Initializer if using Puma in clustered mode
That is, multiple workers configured in puma.rb
:
workers ENV.fetch("WEB_CONCURRENCY") { 2 }
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 = 'https://unleash.herokuapp.com/api'
config.custom_http_headers = {'Authorization': '<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
By not using preload_app!
:
- the
Rails
constant will NOT be available. - but phased restarts will be possible.
You need to ensure that in puma.rb
:
- loading unleash sdk with
require 'unleash'
explicitly, as it will not be pre-loaded. - all parameters must be explicitly set in the
on_worker_boot
block, asconfig/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: 'https://unleash.herokuapp.com/api',
custom_http_headers: {'Authorization': '<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.
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 = 'https://unleash.herokuapp.com/api'
config.custom_http_headers = {'Authorization': '<API token>'}
end
UNLEASH = Unleash::Client.new
end
end
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
Be sure to 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
Or if you see better fit, only in the controllers that you will be using unleash.
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 feature name and context as an 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 boolean via
!!
. - If both a
default_value
andfallback_function
are supplied, the client will define the default value byOR
ing 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
.
If no variant 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')}"
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: 'https://unleash.herokuapp.com/api',
app_name: 'my_ruby_app',
custom_http_headers: { 'Authorization': '<API token>' },
bootstrap_config: Unleash::Bootstrap::Configuration.new({
url: "https://unleash.herokuapp.com/api/client/features",
url_headers: {'Authorization': '<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 though theUNLEASH_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 though theUNLEASH_BOOTSTRAP_URL
environment variable.url_headers
- Headers for the GET http request to theurl
above. Only used if theurl
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 saving the toggles locally:
curl -H 'Authorization: <API token>' -XGET 'https://unleash.herokuapp.com/api' > ./default-toggles.json
Now using them on start up:
custom_boostrapper = lambda {
File.read('./default-toggles.json')
}
@unleash = Unleash::Client.new(
app_name: 'my_ruby_app',
url: 'https://unleash.herokuapp.com/api',
custom_http_headers: { 'Authorization': '<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.
Method Name | Description | Return Type |
---|---|---|
is_enabled? |
Check if feature toggle is to be enabled or not. | Boolean |
enabled? |
Alias to the is_enabled? method. But more ruby idiomatic. |
Boolean |
if_enabled |
Run a code block, if a feature is enabled. | yield |
is_disabled? |
Check if feature toggle is to be enabled or not. | Boolean |
disabled? |
Alias to the is_disabled? method. But more ruby idiomatic. |
Boolean |
if_disabled |
Run a code block, if a feature is disabled. | yield |
get_variant |
Get variant for a given feature | Unleash::Variant |
shutdown |
Save metrics to disk, flush metrics to server, and then kill ToggleFetcher and MetricsReporter threads. A safe shutdown. Not really useful in long running applications, like web applications. | nil |
shutdown! |
Kill ToggleFetcher and MetricsReporter threads immediately. | nil |
For the full method signatures, please see client.rb
# cli unleash client:
bundle exec bin/unleash-client --help
# or a simple sample implementation (with values hardcoded):
bundle exec examples/simple.rb
This client comes with the all the required strategies out of the box:
- ApplicationHostnameStrategy
- DefaultStrategy
- FlexibleRolloutStrategy
- GradualRolloutRandomStrategy
- GradualRolloutSessionIdStrategy
- GradualRolloutUserIdStrategy
- RemoteAddressStrategy
- UnknownStrategy
- UserWithIdStrategy
Client allows to add custom activation strategies using configuration.
In order for 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
After checking out the repo, run bin/setup
to install dependencies.
Then, run 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, rake spec
will also run the client specification tests.
To install this gem onto your local machine, run bundle exec rake install
.
Choose a new version number following Semantic Versioning semantics and then:
- update the version number in ./lib/unleash/version.rb,
- if a major or minor version bump, update the Installation section in README.md
- update CHANGELOG.md following the format on Keep a Changelog
- commit with message
chore: bump version to x.y.z
- then run
bundle exec rake release
- This will create a git tag for the version on the current commit,
- push git commits and tags to origin and
- push the
.gem
file to rubygems.org
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