Flags

Feature toggles service based on dynamic segmentation and criteria.

Reads and writes flags configuration to/from a key-value store (currently Zookeeper) to be used to enable/disable features.

Can be used in multiple domains and applications of your business.

Notions and terminology

Flag

And sometimes also refered to as a Key or as a Feature, it's used to determine whether a certain feature is enabled or disabled.

Future: to be used as a configuration value. e.g. max and min configuration values of certain features, timeouts ... etc.

Application

Each set of features are bound to a certain application domain. This ensures isolation and prevents conflicts of flags in different domains.

Segmentation

A segmentation is how you would like to segment enabling or disabling your features. Can be by the user account type, or by the user device type, or the country, or all of the above. Example segmentations: platform, country.

Segmentation Option

Different types of a given segmentation, say platform segmentation, it has the following possible options: ios, androind, windows, blackberry, pwa, bot ... etc.

Authentication

Currently not supported.

Endpoints

Health checks

• Nginx-level health check:

Endpoint: GET http://{{host}}:{{port}}/check.txt

Expected response:

HTTP/1.0 200 OK

• Application-level health check:

Endpoint: GET http://{{host}}:{{port}}/v1/check

Expected response:

HTTP/1.0 200 OK
Content-Type: text/html; charset=UTF-8

Up and running :)

Read all user-specific features (Basic mode)

This endpoint will return a list of all the features in an application, and whether they're enabled or disabled for the specified user segmentation in the get params.

Endpoint: http://<host>/v1/api/<application>/basic/?<segment>=<option>

Sample response:

{
    "F1": true,
    "F2": true
}

Read all features configuration (Advanced)

This endpoint will return a list of all the features in an application, along with their full configuration, and also a "user_enabled" key which specifies whether a feature enabled or disabled for the specified user segmentation in the get params.

Endpoint: http://<host>/v1/api/<application>/advanced/?<segment>=<option>

Sample response:

{
  "F1": {
    "segmentation": {
      "country": {
        "options": {
          "4": false,
          "7": false
        },
        "toggled": true
      },
      "platform": {
        "options": {
          "android": true,
          "ios": false
        },
        "toggled": true
      }
    },
    "user_enabled": false,
    "feature_toggled": true
  },
  "F2": {
    "segmentation": {
      "country": {
        "options": {
          "4": true,
          "7": true
        },
        "toggled": true
      },
      "platform": {
        "options": {
          "android": true,
          "ios": true
        },
        "toggled": true
      }
    },
    "user_enabled": true,
    "feature_toggled": true
  }
}

Read a user-specific feature (Basic)

Get a single feature in a basic parsed response This endpoint will return whether a feature is enabled or disabled for the specified user segmentation in the get params.

Endpoint: http://<host>/v1/api/<application>/basic/<feature>/?<segment>=<option>

Sample response:

{"F1": true}

Read a single feature's full configuration (Advanced)

This endpoint will return a single feature's full configuration, and also a "user_enabled" key which specifies whether that feature enabled or disabled for the specified user segmentation in the get params.

Endpoint: http://<host>/v1/api/<application>/advanced/<feature>/?<segment>=<option>

Sample response:

{
  "segmentation": {
    "country": {
      "options": {
        "4": false,
        "7": false
      },
      "toggled": true
    },
    "platform": {
      "options": {
        "android": true,
        "ios": false
      },
      "toggled": true
    }
  },
  "user_enabled": true,
  "feature_toggled": true
}

Setup

You can setup the application using Docker:

• Install Docker
• After you clone the repo, inside the repo dir, run make docker This will take care of all the setup, package installation and configuration needed for the application to run.
• Then after you build the docker container, run it providing your specific settings:

docker run -p 8095:80 -e NEW_RELIC_LICENSE_KEY="fake-key" -e PRODUCTION_MODE="0" -e SYSLOG_HOST="your-syslog-host" -e SYSLOG_PORT="your-syslog-port" -e VENV_DIR="python-virtualenv-dir-path" e REPO_DIR="dir-repo-path" -e APP_NAME="flags" -e ZK_HOSTS="comma-separated-ZK-hosts-and-ports" your-docker-image-here

Monitoring

The application is integrated with NewRelic by default, just provide the account license key as a docker env var NEW_RELIC_LICENSE_KEY when you run the docker container, and enable PRODUCTION_MODE:

-e NEW_RELIC_LICENSE_KEY="your_license_key_here" -e PRODUCTION_MODE="1"

What's next:

list of improvements