unleash.unleash/website/docs/sdks/unleash-proxy.md

id	title
unleash-proxy	Unleash Proxy

The unleash-proxy is compatible with all Unleash Enterprise versions and Unleash Open-Source v4. You should reach out to support@getunleash.io if you want the Unleash Team to host the Unleash Proxy for you.

A lot of our users wanted to use feature toggles in their single-page and native applications. To solve this in a performant and privacy concerned way we built The Unleash Proxy

The Unleash Proxy sits between the Unleash API and the application. It provides a simple and super-fast API, as it has all the data it needs available in memory.

The proxy solves three important aspects:

• Performance – The proxy will cache all toggles in memory, and will be running on the edge, close to your end-users. A single instance will be able to handle thousands of request/sec, and you can scale it easily by adding additional instances.
• Security – The proxy evaluates the feature flags for the user on the server-side, and exposes results for feature flags that are enabled for a specific user (flags not enabled for that specific user are not exposed).
• Privacy – If you run the proxy yourself (we can host it as well though) we will not see your end users. This means that you still have full control of your end-users, the way it should be!

The Unleash Proxy uses the Unleash SDK and exposes a simple API. The Proxy will synchronize with the Unleash API in the background and provide a simple HTTP API for clients.

Configuration

Required configuration variables

Regardless of how you choose to run the it, the proxy will need access to these three variables:

• unleashUrl / UNLEASH_URL

  The URL of your Unleash instance's API. For instance, to connect to the Unleash demo app, you would use https://app.unleash-hosted.com/demo/api/

• unleashApiToken / UNLEASH_API_TOKEN

  The API token to connect to your Unleash project. For more information on how these work and how to create them, check out the API token documentation.

• clientKeys / UNLEASH_PROXY_CLIENT_KEYS

  A list of client keys that the proxy will accept. For the proxy to accept an incoming request, the client must use one of these keys for authorization. In client SDKs, this is usually known as a clientKey or a clientSecret. If you query the proxy directly via HTTP, this is the authorization header.

  When using an environment variable to set the proxy secrets, the value should be a comma-separated list of strings, such as secret-one,secret-two.

There are many more configuration options available. You'll find all available options on github.

All configuration options

Option	Environment Variable	Default value	Required	Description
clientKeys	UNLEASH_PROXY_CLIENT_KEYS	n/a	yes	List of client keys that the proxy should accept. When querying the proxy, Proxy SDKs must set the request's client keys header to one of these values. The default client keys header is Authorization.
clientKeysHeaderName	CLIENT_KEY_HEADER_NAME	"authorization"	no	The name of the HTTP header to use for client keys. Incoming requests must set the value of this header to one of the Proxy's clientKeys to be authorized successfully.
customStrategies	UNLEASH_CUSTOM_STRATEGIES_FILE	[]	no	Use this option to inject implementation of custom activation strategies. If you are using UNLEASH_CUSTOM_STRATEGIES_FILE you need to provide a valid path to a javascript files which exports an array of custom activation strategies and the SDK will automatically load these
environment	UNLEASH_ENVIRONMENT	undefined	no	If set this will be the environment used by the proxy in the Unleash Context. It will not be possible for proxy SDKs to override the environment if set.
logLevel	LOG_LEVEL	"warn"	no	Used to set logLevel. Supported options: "debug", "info", "warn", "error" and "fatal"
logger	n/a	SimpleLogger	no	Register a custom logger.
metricsInterval	UNLEASH_METRICS_INTERVAL	30000	no	How often the proxy should send usage metrics back to Unleash, defined in ms.
namePrefix	UNLEASH_NAME_PREFIX	undefined	no	Used to filter features by using prefix when requesting backend values.
projectName	UNLEASH_PROJECT_NAME	undefined	no	The projectName (id) to fetch feature toggles for. The proxy will only return know about feature toggles that belongs to the project, if specified.
proxyBasePath	PROXY_BASE_PATH	""	no	The base path to run the proxy from. "/proxy" will be added at the end. For instance, if proxyBasePath is "base/path", the proxy will run at /base/path/proxy.
proxyPort	PORT	3000	no	The port where the proxy should listen.
proxySecrets	UNLEASH_PROXY_SECRETS	n/a	no	Deprecated alias for clientKeys. Please use clientKeys instead.
refreshInterval	UNLEASH_FETCH_INTERVAL	5000	no	How often the proxy should query Unleash for updates, defined in ms.
tags	UNLEASH_TAGS	undefined	no	Used to filter features by using tags set for features. Format should be tagName:tagValue,tagName2:tagValue2
trustProxy	TRUST_PROXY	false	no	By enabling the trustProxy option, Unleash Proxy will have knowledge that it's sitting behind a proxy and that the X-Forwarded-* header fields may be trusted, which otherwise may be easily spoofed. The proxy will automatically enrich the ip address in the Unleash Context. Can either be true/false (Trust all proxies), trust only given IP/CIDR (e.g. '127.0.0.1') as a string. May be a list of comma separated values (e.g. '127.0.0.1,192.168.1.1/24'
unleashApiToken	UNLEASH_API_TOKEN	n/a	yes	API token (client) needed to connect to Unleash API.
unleashAppName	UNLEASH_APP_NAME	"unleash-proxy"	no	App name to used when registering with Unleash
unleashInstanceId	UNLEASH_INSTANCE_ID	generated	no	Unleash instance id to used when registering with Unleash
unleashUrl	UNLEASH_URL	n/a	yes	API Url to the Unleash instance to connect to

Health endpoint

The proxy will try to synchronize with the Unleash API at startup, until it has successfully done that the proxy will return HTTP 503 - Not Read? for all request. You can use the health endpoint to validate that the proxy is ready to recieve requests:

curl http://localhost:3000/proxy/health -I

HTTP/1.1 200 OK
Access-Control-Allow-Origin: *
Access-Control-Expose-Headers: ETag
Content-Type: text/html; charset=utf-8
Content-Length: 2
ETag: W/"2-eoX0dku9ba8cNUXvu/DyeabcC+s"
Date: Fri, 04 Jun 2021 10:38:27 GMT
Connection: keep-alive
Keep-Alive: timeout=5

Custom activation strategies

The Unleash Proxy can load custom activation strategies for front-end client SDKs (Android, JavaScript, React, iOS). For a step-by-step guide, refer to the how to use custom strategies guide.

To load custom strategies, use either of these two options:

• the customStrategies option: use this if you're running the Unleash Proxy via Node directly.
• the UNLEASH_CUSTOM_STRATEGIES_FILE environment variable: use this if you're running the proxy as a container.

Both options take a list of file paths to JavaScript files that export custom strategy implementations.

Custom activation strategy files format

Each strategy file must export a list of instantiated strategies. A file can export as many strategies as you'd like.

Here's an example file that exports two custom strategies:

const { Strategy } = require('unleash-client');

class MyCustomStrategy extends Strategy {
  // ... strategy implementation
}

class MyOtherCustomStrategy extends Strategy {
  // ... strategy implementation
}

// export strategies
module.exports = [
    new MyCustomStrategy(),
    new MyOtherCustomStrategy()
];

Refer the custom activation strategy documentation for more details on how to implement a custom activation strategy.

Unleash Proxy API

The Unleash Proxy has a very simple API. It takes the Unleash Context as input and will return the feature toggles relevant for that specific context.

Payload

The proxy endpoint returns information about toggles enabled for the current user. The payload is a JSON object with a toggles property, which contains a list of toggles.

{
  "toggles": [
    {
      "name": "demo",
      "enabled": true,
      "variant": {
        "name": "disabled",
        "enabled": false
      }
    },
    {
      "name": "demoApp.step1",
      "enabled": true,
      "variant": {
        "name": "disabled",
        "enabled": false
      }
    }
  ]
}

Toggle data

The data for a toggle without variants looks like this:

{
  "name": "basic-toggle",
  "enabled": true,
  "variant": {
    "name": "disabled",
    "enabled": false
  }
}

• name: the name of the feature.
• enabled: whether the toggle is enabled or not. Will always be true.
• variant: describes whether the toggle has variants and, if it does, what variant is active for this user. If a toggle doesn't have any variants, it will always be {"name": "disabled", "enabled": false}.

:::note Unleash uses a fallback variant called "disabled" to indicate that a toggle has no variants. However, you are free to create a variant called "disabled" yourself. In that case you can tell them apart by checking the variant's enabled property: if the toggle has no variants, enabled will be false. If the toggle is the "disabled" variant that you created, it will have enabled set to true. :::

If a toggle has variants, then the variant object can also contain an optional payload property. The payload will contain data about the variant's payload: what type it is, and what the content is. To learn more about variants and their payloads, check the feature toggle variants documentation.

Variant toggles without payloads look will have their name listed and the enabled property set to true:

{
  "name": "toggle-with-variants",
  "enabled": true,
  "variant": {
    "name": "simple",
    "enabled": true
  }
}

If the variant has a payload, the optional payload property will list the payload's type and it's content in a stringified form:

{
  "name": "toggle-with-variants",
  "enabled": true,
  "variant": {
    "name": "with-payload-string",
    "payload": {
      "type": "string",
      "value": "this string is the variant's payload"
    },
    "enabled": true
  }
}

For the variant property:

• name: is the name of the variant, as shown in the Admin UI.
• enabled: indicates whether the variant is enabled or not. If the toggle has variants, this is always true.
• payload (optional): Only present if the variant has a payload. Describes the payload's type and content.

If the variant has a payload, the payload object contains:

• type: the type of the variant's payload
• value: the value of the variant's payload

The value will always be the payload's content as a string, escaped as necessary. For instance, a variant with a JSON payload would look like this:

{
  "name": "toggle-with-variants",
  "enabled": true,
  "variant": {
    "name": "with-payload-json",
    "payload": {
      "type": "json",
      "value": "{\"description\": \"this is data delivered as a json string\"}"
    },
    "enabled": true
  }
}

We care about Privacy!

The Unleash Proxy is important because you should not expose your entire set of toggle configurations to your end users. Single page apps work in the context of a specific user. The proxy allows you to only provide data that relates to that one user: The proxy will only return the evaluated toggles (with variants) that should be enabled for that specific user in that specific context.

Most of our customers prefer to run The Unleash Proxy themselves. PS! We actually prefer this as we don’t want to see your users. Running it is pretty simple, it is either a small Node.js process you start or a docker image you use. (We can of course host the proxy for you also.)

How to connect to the Proxy?

The Unleash Proxy takes the heavy lifting of evaluating toggles and only returns enabled toggles and their values to the client. This means that you would get away with a simple http-client in many common use-cases.

However in some settings you would like a bit more logic around it to make it as fast as possible, and keep up to date with changes.

• JavaScript Proxy SDK
• React Proxy SDK
• Android Proxy SDK
• iOS Proxy SDK

The proxy is also ideal fit for serverless functions such as AWS Lambda. In that scenario the proxy can run on a small container near the serverless function, preferably in the same VPC, giving the lambda extremely fast access to feature flags, at a predictable cost.

Configuration options

Regardless of how you choose to run the it, the proxy will need access to these three variables:

• unleashUrl / UNLEASH_URL

  The URL of your Unleash instance's API. For instance, to connect to the Unleash demo app, you would use https://app.unleash-hosted.com/demo/api/

• unleashApiToken / UNLEASH_API_TOKEN

  The API token to connect to your Unleash project. For more information on how these work and how to create them, check out the API token documentation.

• clientKeys / UNLEASH_PROXY_CLIENT_KEYS

  A list of client keys that the proxy will accept. For the proxy to accept an incoming request, the client must use one of these keys for authorization. In client SDKs, this is usually known as a clientKey or a clientSecret. If you query the proxy directly via HTTP, this is the authorization header.

  When using an environment variable to set the proxy secrets, the value should be a comma-separated list of strings, such as secret-one,secret-two.

There are many more configuration options available. You'll find all available options on github.

Bootstrapping the Unleash Proxy

Scaling the Unleash Proxy

Hosting the Unleash Proxy

Proxy connection

{ how to connect the proxy, what setups are possible etc. } See #bootstrap

Configuration options