unleash.unleash/website/docs/using-unleash/deploy/configuring-unleash.md

title
Configuring Unleash

import Tabs from '@theme/Tabs'; import TabItem from '@theme/TabItem';

This is the guide on how to configure Unleash v5 self-hosted.

Must configure

Database

In order for Unleash server to work, you need a running database and its connection details. See the database configuration section for the available options and configuration details.

Nice to configure

Unleash URL

• Configured with UNLEASH_URL ** Should be set to the public discoverable URL of your instance, so if your instance is accessed by your users at https://unleash.mysite.com use that. ** If you're deploying this to a subpath, include the subpath in this. So https://mysite.com/unleash will also be correct.
• Used to create
  • Reset password URLs
  • Welcome link for new users
  • Links in events for our Slack, Microsoft Teams and Datadog integrations

Email server details

Used to send reset-password mails and welcome mails when onboarding new users.
NOTE - If this is not configured, you will not be able to allow your users to reset their own passwords.

For more details, see here

Minimum and recommended specifications

Minimum

Resource	Request	Limit
CPU	250m	500m
Memory	200Mi	400Mi

Recommended

Resource	Request	Limit
CPU	500m	1000m
Memory	512Mi	1Gi

Further customization

In order to customize "anything" in Unleash you need to use Unleash from Node.js or start the docker image with environment variables.

const unleash = require('unleash-server');

const unleashOptions = {
    db: {
        user: 'unleash_user',
        password: 'password',
        host: 'localhost',
        port: 5432,
        database: 'unleash',
        ssl: false,
        pool: {
            min: 0,
            max: 4,
            idleTimeoutMillis: 30000,
        },
    },
    enableRequestLogger: true,
};

unleash.start(unleashOptions);

Available Unleash options include:

• authentication - (object) - An object for configuring/implementing custom admin authentication

  • enableApiToken / AUTH_ENABLE_API_TOKEN: boolean — Should unleash require API tokens for access? Defaults to true.

  • type / AUTH_TYPE: string — What kind of authentication to use. Possible values

    • open-source - Sign in with username and password. This is the default value.
    • custom - If implementing your own authentication hook, use this
    • demo - Only requires an email to sign in (was default in v3)
    • none - Deprecated Turn off authentication completely. If no API token is provided towards /api/client or /api/frontend you will receive configuration for the "default" environment. We generally recommend you use the demo type for simple, insecure usage of Unleash. This auth type has many known limitations, particularly related to personalized capabilities such as favorites and notifications.

  • customAuthHandler: function (app: any, config: IUnleashConfig): void — custom express middleware handling authentication. Used when type is set to custom. Can not be set via environment variables.

  • initialAdminUser: { username: string, password: string} | null — whether to create an admin user with default password - Defaults to using admin and unleash4all as the username and password. Can not be overridden by setting the UNLEASH_DEFAULT_ADMIN_USERNAME and UNLEASH_DEFAULT_ADMIN_PASSWORD environment variables.

  • initApiTokens / INIT_ADMIN_API_TOKENS, INIT_CLIENT_API_TOKENS, and INIT_FRONTEND_API_TOKENS: ApiTokens[] — Array of API tokens to create on startup. The tokens will only be created if the database doesn't already contain any API tokens. Example:

    [
      {
        environment: '*',
        project: '*',
        secret: '*:*.964a287e1b728cb5f4f3e0120df92cb5',
        type: ApiTokenType.ADMIN,
        username: 'some-user',
      },
    ];
    

    The tokens can be of any API token type. Note that admin tokens must target all environments and projects ( i.e. use '*' for environments and project and start the secret with *:*.).

    You can also use the environment variables INIT_ADMIN_API_TOKENS, INIT_CLIENT_API_TOKENS, and INIT_FRONTEND_API_TOKENS. All variables require a comma-separated list of API tokens to initialize (for instance *:*.some-random-string, *:*.some-other-token). The tokens found in INIT_ADMIN_API_TOKENS, INIT_CLIENT_API_TOKENS, and INIT_FRONTEND_API_TOKENS will be created as admin, client, and frontend tokens respectively, and Unleash will assign usernames automatically.

  • demoAllowAdminLogin / AUTH_DEMO_ALLOW_ADMIN_LOGIN: boolean — Allows you to log in as the admin user when running Unleash with the demo auth type. To log in as the admin user, use admin as your email. Defaults to false.

• databaseUrl - (deprecated) the postgres database url to connect to. Only used if db object is not specified, and overrides the db object and any environment variables that change parts of it (like DATABASE_SSL). Should include username/password. This value may also be set via the DATABASE_URL environment variable. Alternatively, if you would like to read the database url from a file, you may set the DATABASE_URL_FILE environment variable with the full file path. The contents of the file must be the database url exactly.

• db - The database configuration object. See the database configuration section for a full overview of the available properties.

• disableLegacyFeaturesApi (boolean) - whether to disable the legacy features API. Defaults to false (DISABLE_LEGACY_FEATURES_API). Introduced in Unleash 4.6.

• email - the email object configuring an SMTP server for sending welcome mails and password reset mails

  • host - The server URL to your SMTP server
  • port - Which port the SMTP server is running on. Defaults to 465 (Secure SMTP)
  • secure (boolean) - Whether to use SMTPS or not.
  • sender - Which email should be set as sender of mails being sent from Unleash?
  • smtpuser - Username for your SMTP server
  • smtppass - Password for your SMTP server

• eventHook (function(event, data)) - (deprecated in Unleash 4.3 in favor of the Webhook integration. Removed in Unleash 5) If provided, this function will be invoked whenever a feature is mutated. The possible values for event are 'feature-created', 'feature-archived' and 'feature-revived'. The data argument contains information about the mutation. Its fields are type (string) - the event type (same as event); createdBy (string) - the user who performed the mutation; data - the contents of the change. The contents in data differs based on the event type; For 'feature-archived' and 'feature-revived', the only field will be name - the name of the feature. For 'feature-created' the data follows a schema defined in the code here. See an api here.

• getLogger (function) - Used to register a custom log provider.

• logLevel (debug | info | warn | error | fatal) - The lowest level to log at, also configurable using environment variable LOG_LEVEL.

• enableRequestLogger (boolean) - use this to enable logging for requested urls and response codes (default: false).

• preHook (function) - this is a hook if you need to provide any middlewares to express before unleash adds any. Express app instance is injected as first argument.

• preRouterHook (function) - use this to register custom express middlewares before the unleash specific routers are added.

• secureHeaders (boolean) - use this to enable security headers (HSTS, CSP, etc) when serving Unleash from HTTPS. Can also be configured through the environment variable SECURE_HEADERS.

• additionalCspAllowedDomains (CspAllowedDomains) - use this when you want to enable security headers but have additional domains you need to allow traffic to. You can set the following environment variables:

  • CSP_ALLOWED_DEFAULT to allow new defaultSrc (comma separated list)
  • CSP_ALLOWED_FONT to allow new fontSrc (comma separated list)
  • CSP_ALLOWED_STYLE to allow new styleSrc (comma separated list)
  • CSP_ALLOWED_SCRIPT to allow new scriptSrc (comma separated list)
  • CSP_ALLOWED_IMG to allow new imgSrc (comma separated list)
  • CSP_ALLOWED_CONNECT to allow new connectSrc (comma separated list)
  • CSP_ALLOWED_FRAME to allow new frameSrc (comma separated listed)
  • CSP_ALLOWED_MEDIA to allow new mediaSrc (comma separated list)
  • CSP_ALLOWED_OBJECT to allow new objectSrc (comma separated list)

• server - The server config object taking the following properties

  • port - which port the unleash-server should bind to. If port is omitted or is 0, the operating system will assign an arbitrary unused port. Will be ignored if pipe is specified. This value may also be set via the HTTP_PORT environment variable
  • host - which host the unleash-server should bind to. If host is omitted, the server will accept connections on the unspecified IPv6 address (::) when IPv6 is available, or the unspecified IPv4 address (0.0.0.0) otherwise. This value may also be set via the HTTP_HOST environment variable
  • pipe - parameter to identify IPC endpoints. See https://nodejs.org/api/net.html#net_identifying_paths_for_ipc_connections for more details
  • serverMetrics (boolean) - use this option to turn on/off prometheus metrics.
  • baseUriPath (string) - use to register a base path for all routes on the application. For example /my/unleash/base (note the starting /). Defaults to /. Can also be configured through the environment variable BASE_URI_PATH.
  • unleashUrl (string) - Used to specify the official URL this instance of Unleash can be accessed at for an end user. Can also be configured through the environment variable UNLEASH_URL.
  • gracefulShutdownEnable: (boolean) - Used to control if Unleash should shutdown gracefully (close connections, stop tasks,). Defaults to true. GRACEFUL_SHUTDOWN_ENABLE
  • gracefulShutdownTimeout: (number) - Used to control the timeout, in milliseconds, for shutdown Unleash gracefully. Will kill all connections regardless if this timeout is exceeded. Defaults to 1000ms GRACEFUL_SHUTDOWN_TIMEOUT
  • enableScheduledCreatedByMigration: (boolean) - Schedules migrations that fills the field created_by_user_id for past events and features. It does it gradually in chunks and runs every 15 minutes so it doesn't affect the application's performance. Defaults to false. NOTE: If your installation started with Unleash v5.8 or later, you should not need this flag! ENABLE_SCHEDULED_CREATED_BY_MIGRATION

• session - The session config object takes the following options:

  • ttlHours (number) - The number of hours a user session is allowed to live before a new sign-in is required. Defaults to 48 hours. SESSION_TTL_HOURS
  • clearSiteDataOnLogout (boolean) - When true, a logout action will return a Clear Site Data response header instructing the browser to clear all cookies on the same domain Unleash is running on. If disabled unleash will only destroy and clear the session cookie. Defaults to true. SESSION_CLEAR_SITE_DATA_ON_LOGOUT
  • cookieName - Name of the cookies used to hold the session id. Defaults to 'unleash-session'.

• ui (object) - Set of UI specific overrides. You may set the following keys: environment, slogan.

• versionCheck - the object deciding where to check for latest version

  • url - The url to check version (Defaults to https://version.unleash.run) - Overridable with (UNLEASH_VERSION_URL)
  • enable - Whether version checking is enabled (defaults to true) - Overridable with (CHECK_VERSION) (if anything other than true, does not check)

• environmentEnableOverrides - A list of environment names to force enable at startup. This is feature should be used with caution. When passed a list, this will enable each environment in that list and disable all other environments. You can't use this to disable all environments, passing an empty list will do nothing. If one of the given environments is not already enabled on startup then it will also enable projects and toggles for that environment. Note that if one of the passed environments doesn't already exist this will do nothing aside from log a warning.

• clientFeatureCaching - configuring memoization of the /api/client/features endpoint

  • enabled - set to true by default - Overridable with (CLIENT_FEATURE_CACHING_ENABLED)
  • maxAge - the time to cache features, set to 600 milliseconds by default - Overridable with (CLIENT_FEATURE_CACHING_MAXAGE) ) (accepts milliseconds)

• frontendApi - Configuration options for the Unleash front-end API.

  • refreshIntervalInMs - how often (in milliseconds) front-end clients should refresh their data from the cache. Overridable with the FRONTEND_API_REFRESH_INTERVAL_MS environment variable.

• accessControlMaxAge - You can configure the max-age of the Access-Control-Max-Age header. Defaults to 86400 seconds. Overridable with the ACCESS_CONTROL_MAX_AGE environment variable.

• responseTimeWithAppNameKillSwitch - use this to disable metrics with app names. This is enabled by default but may increase the cardinality of metrics causing Unleash memory usage to grow if your app name is randomly generated (which is not recommended). Overridable with the UNLEASH_RESPONSE_TIME_WITH_APP_NAME_KILL_SWITCH environment variable.

• disableCompression - Disables Express compression middleware, useful when configuring the application in a serverless environment. Defaults to false. Overridable with the SERVER_DISABLE_COMPRESSION environment variable.

• keepAliveTimeout - Use this to tweak connection keepalive timeout in seconds. Useful for hosted situations where you need to make sure your connections are closed before terminating the instance. Defaults to 15. Overridable with the SERVER_KEEPALIVE_TIMEOUT environment variable. You can also set the environment variable ENABLED_ENVIRONMENTS to a comma delimited string of environment names to override environments.

• metricsRateLimiting - Use the following to tweak the rate limits for /api/client/register, /api/client/metrics, /api/frontend/register and /api/frontend/metrics POST endpoints

  • clientMetricsMaxPerMinute - How many requests per minute per IP is allowed against POST /api/client/metrics before returning 429. Set to 6000 by default (100rps) - Overridable with REGISTER_CLIENT_RATE_LIMIT_PER_MINUTE environment variable
  • clientRegisterMaxPerMinute - How many requests per minute per IP is allowed against POST /api/client/register before returning 429. Set to 6000 by default (100rps) - Overridable with CLIENT_METRICS_RATE_LIMIT_PER_MINUTE environment variable
  • frontendMetricsMaxPerMinute - How many requests per minute per IP is allowed against POST /api/frontend/metrics before returning 429. Set to 6000 by default (100rps) - Overridable with FRONTEND_METRICS_RATE_LIMIT_PER_MINUTE environment variable
  • frontendRegisterMaxPerMinute - How many requests per minute per IP is allowed against POST /api/frontend/register before returning 429. Set to 6000 by default (100rps) - Overridable with REGISTER_FRONTEND_RATE_LIMIT_PER_MINUTE environment variable

• rateLimiting - Use the following to tweak the rate limits for POST /auth/simple (Password-based login) and POST /api/admin/user-admin (Creating users)

  • simpleLoginMaxPerMinute - How many requests per minute per IP is allowed against POST /auth/simple before returning 429. Set to 10 by default - Overridable with SIMPLE_LOGIN_LIMIT_PER_MINUTE environment variable
  • createUserMaxPerMinute - How many requests per minute per IP is allowed against POST /api/admin/user-admin before returning 429. Set to 20 by default - Overridable with CREATE_USER_RATE_LIMIT_PER_MINUTE environment variable

Disabling Auto-Start

If you're using Unleash as part of a larger express app, you can disable the automatic server start by calling server.create. It takes the same options as server.start, but will not begin listening for connections.

const express = require('express');
const unleash = require('unleash-server');
const app = express();

const start = async () => {
    const instance = await unleash.create({
        databaseUrl: 'postgres://unleash_user:password@localhost:5432/unleash',
    });
    app.use(instance.app);
    console.log(`Unleash app generated and attached to parent application`);
};

start();

Graceful shutdown

PS! Unleash will listen for the SIGINT signal and automatically trigger graceful shutdown of Unleash.

If you need to stop Unleash (close database connections, and stop running Unleash tasks) you may use the stop function. Be aware that it is not possible to restart the Unleash instance after stopping it, but you can create a new instance of Unleash.

const express = require('express');
const unleash = require('unleash-server');
const app = express();

const start = async () => {
    const instance = await unleash.start({
        databaseUrl: 'postgres://unleash_user:password@localhost:5432/unleash',
        port: 4242,
    });

    //Sometime later
    await instance.stop();
    console.log('Unleash has now stopped');
};

start();

Segment limits

:::caution

Changing segment limits could have a negative impact on the performance of Unleash SDKs and cause network congestion. Think twice before changing these values.

:::

Some facets of the segments feature can be customized via environment variables. This lets you change the segment limits that Unleash uses.

UNLEASH_STRATEGY_SEGMENTS_LIMIT controls the maximum number of segments that can be applied to a single strategy. The default is 5.

UNLEASH_SEGMENT_VALUES_LIMIT controls the maximum number of values that you can assign across a segment's constraints. The default is 1000 (Since v5.1.0; it was 100 before v5.1.0)

Securing Unleash

You can integrate Unleash with your authentication provider (OAuth 2.0). Read more about securing unleash.

How do I configure the log output?

By default, unleash uses log4js to log important information. It is possible to swap out the logger provider (only when using Unleash programmatically). You do this by providing an implementation of the getLogger function as This enables filtering of log levels and easy redirection of output streams.

function getLogger(name) {
    // do something with the name
    return {
        debug: console.log,
        info: console.log,
        warn: console.log,
        error: console.error,
    };
}

The logger interface with its debug, info, warn and error methods expects format string support as seen in debug or the JavaScript console object. Many commonly used logging implementations cover this API, e.g., bunyan, pino or winston.

Database configuration

:::info

In-code configuration values take precedence over environment values: If you provide a database username both via environment variables and in code with the Unleash options object, Unleash will use the in-code username.

:::

You cannot run the Unleash server without a database. You must provide Unleash with database connection details for it to start correctly.

The available options are listed in the table below. Options can be specified either via JavaScript (only when starting Unleash via code) or via environment variables. The "property name" column below gives the name of the property on the Unleash options' db object.

Property name	Environment variable	Default value	Description
user	DATABASE_USERNAME	unleash_user	The database username.
password	DATABASE_PASSWORD	password	The database password.
host	DATABASE_HOST	localhost	The database hostname.
port	DATABASE_PORT	5432	The database port.
database	DATABASE_NAME	unleash	The name of the database.
ssl	DATABASE_SSL	N/A	An object describing SSL options. In code, provide a regular JavaScript object. When using the environment variable, provide a stringified JSON object.
	DATABASE_SSL_CA_CONFIG	N/A	JSON file representing the SSL options, needs 4 keys set, rejectUnauthorized, ca, cert, key. the ca, cert and key keys needs to be ssl certificates as a string.
	DATABASE_SSL_CA_FILE	N/A	A path to a valid SSL CA file in pem format. Sets the ca key of the ssl object. See [https://node-postgres.com/features/ssl](pg docs)
	DATABASE_SSL_CERT_FILE	N/A	A path to a valid SSL Cert file in pem format. Sets the cert key of the ssl object. See [https://node-postgres.com/features/ssl](pg docs)
	DATABASE_SSL_KEY_FILE	N/A	A path to a valid SSL key file in pem format. Sets the key key of the ssl object. See [https://node-postgres.com/features/ssl](pg docs)
	DATABASE_SSL_REJECT_UNAUTHORIZED	true if CA, CERT and KEY is set, false otherwise	If you use SSL connections to postgres, don't set this to false. It will turn off the security check of SSL. See [https://node-postgres.com/features/ssl](pg docs)
pool	N/A (use the below variables)		An object describing database pool options. With environment variables, use the options below directly.
pool.min	DATABASE_POOL_MIN	0	The minimum number of connections in the connection pool.
pool.max	DATABASE_POOL_MAX	4	The maximum number of connections in the connection pool.
pool.idleTimeoutMillis	DATABASE_POOL_IDLE_TIMEOUT_MS	30000	The amount of time (in milliseconds) that a connection must be idle for before it is marked as a candidate for eviction.
applicationName	DATABASE_APPLICATION_NAME	unleash	The name of the application that created this Client instance.
schema	DATABASE_SCHEMA	public	The schema to use in the database.

Alternatively, you can use a single-host libpq connection string to connect to the database. You can provide it directly or from a file by using one of the below options. In JavaScript, these are top-level properties of the root configuration object, not the db object.

Property name	Environment variable	Default value	Description
databaseUrl	DATABASE_URL	N/A	A string that matches a single-host libpq connection string, such as postgres://USER:PASSWORD@HOST:PORT/DATABASE. Unleash does not support using multiple hosts.
databaseUrlFile	DATABASE_URL_FILE	N/A	The path to a file that contains a libpq connection string.

Below is an example JavaScript configuration object.

const unleashOptions = {
    databaseUrl: 'postgres:/USER:PASSWORD@HOST:PORT/DATABASE',
    databaseUrlFile: '/path/to/file',
    db: {
        user: 'unleash_user',
        password: 'password',
        host: 'localhost',
        port: 5432,
        database: 'unleash',
        ssl: false,
        pool: {
            min: 0,
            max: 4,
            idleTimeoutMillis: 30000,
        },
    },
};

db.ssl vs DATABASE_SSL options

When initializing Unleash from code, you'll provide the db.ssl option as a JavaScript object. As such, any functions will get evaluated before the object is used for configuration. When using the DATABASE_SSL environment variable, you must provide the value as a stringified JSON object. The object will get parsed before being used for configuration, but no further evaluation will take place.

If you want to read content from a file, you should either initialize Unleash via JavaScript or manually interpolate the required values into the environment variable:

const unleashOptions = {
    db: {
        // other options omitted for brevity
        ssl: {
            ca: fs.readFileSync('/path/to/server-certificates/root.crt').toString(),
        },
    },
};

DATABASE_SSL="{ \"ca\": \"$(cat /path/to/server-certificates/root.crt)\" }"

Enabling self-signed certificates

To use self-signed certificates, you should set the SSL property rejectUnauthorized to false and set the ca property to the value of the certificate:

const unleashOptions = {
    db: {
        // other options omitted for brevity
        ssl: {
            rejectUnauthorized: false,
            ca: fs.readFileSync('/path/to/server-certificates/root.crt').toString(),
        },
    },
};

DATABASE_SSL="{ \"rejectUnauthorized\": false, \"ca\": \"$(cat /path/to/server-certificates/root.crt)\" }"

Visit the node-postgres library's SSL section for more information.

Supported Postgres SSL modes

Unleash builds directly on the node-postgres library, so we support all the same SSL modes that they support. As of version 8 (released on February 25th 2020), node-postgres no longer supports all sslmodes. For this reason, Unleash cannot support all of Postgres' SSL modes. Specifically, Unleash does not support sslmode=prefer. Instead you must know whether you require an SSL connection ahead of time.

Recommended PostgreSQL spec

Minimum

For PostgreSQL we recommend running with at least 1 CPU and at least 1Gi of memory, as well as at least 5Gi of storage.

Provider	Machine
AWS	db.t4g.medium (2CPU / 4GB mem)
Azure	B2s (2CPU / 4GB mem )
GCP	2 CPU / 4GB mem

Recommended

PostgreSQL performance improves the more resources you throw at it, and if you have many connections active, you want more RAM, so a good recommendation is:

• 4CPUs and 16GB of memory, and at least 20Gi of SSD storage

Provider	Machine
AWS-RDS	db.m6g.xlarge (4CPU / 16GB mem) or db.m5.xlarge (4cores / 16GB mem)
Azure	Standard_E2bds_v5 (2CPU / 16GB mem )
GCP	4 CPU / 16GB mem

Troubleshooting: database pooling connection timeouts

• Check the default values of connection pool about idle session handling.

• If you have a network component which closes idle sessions on the TCP layer, make sure that the connection pool's idleTimeoutMillis setting is lower than the timespan setting. If it isn't, then the network component will close the connection.

Troubleshooting: missing metrics from connected Edge instances

If you're on Unleash 5.9.0 or newer, you might encounter issues with metrics from connected Edge instances. There are two solutions to this, the recommended way and the one that will work for 5.9.0 but probably not for later versions

Recommended

• Make sure you're using the latest Edge, at least v17.0.0
• Make sure you have at least one valid client token configured. Either using the --token flag on startup, or via connecting an SDK with a valid token

Works in 5.9.0

• If you absolutely cannot upgrade your Edge instances to v17.0.0 or newer you can start Unleash with the environmental flag UNLEASH_EXPERIMENTAL_EDGE_BULK_METRICS set to true
• This will re-enable an endpoint that was disabled in 5.9.0 to allow users that cannot upgrade Edge just now to stay on an old Edge version, but still get client metrics.

Proxying requests from Unleash

You can configure proxy services that intercept all outgoing requests from Unleash. This lets you use the Microsoft Teams or the Webhook integration for external services, even if the internet can only be reached via a proxy on your machine or container (for example if restricted by a firewall or similiar).

As an example, here's how you could do it using the node-global-proxy package:

const proxy = require("node-global-proxy").default;

proxy.setConfig({
    http: "http://user:password@url:8080",      //proxy adress, replace values as needed
    //https: "https://user:password@url:1080",  //if a https proxy is needed
  });

proxy.start();      //this starts the proxy, after this call all requests will be proxied

Using above code-snippet, every outgoing request from unleash or its integrations will be subsequently routed through set proxy. If the proxy routing needs to be bypassed or stopped, its possible to stop it by using

proxy.stop();