Configure Real User Monitoring (RUM)edit

This documentation refers to configuring the standalone (legacy) APM Server. This method of running APM Server will be deprecated and removed in a future release. Please consider upgrading to Fleet and the APM integration. If you’re using Fleet and the Elastic APM integration, please see Real User Monitoring (RUM) instead.

The Real User Monitoring (RUM) agent captures user interactions with clients such as web browsers. These interactions are sent as events to the APM Server. Because the RUM agent runs on the client side, the connection between agent and server is unauthenticated. As a security precaution, RUM is therefore disabled by default. To enable it, set apm-server.rum.enabled to true in your APM Server configuration file.

In addition, if APM Server is deployed in an origin different than the page’s origin, you will need to configure Cross-Origin Resource Sharing (CORS) in the Agent.

Example config with RUM enabled:

apm-server.rum.enabled: true
apm-server.auth.anonymous.rate_limit.event_limit: 300
apm-server.auth.anonymous.rate_limit.ip_limit: 1000
apm-server.auth.anonymous.allow_service: [your_service_name]
apm-server.rum.allow_origins: ['*']
apm-server.rum.allow_headers: ["header1", "header2"]
apm-server.rum.library_pattern: "node_modules|bower_components|~"
apm-server.rum.exclude_from_grouping: "^/webpack"
apm-server.rum.source_mapping.enabled: true
apm-server.rum.source_mapping.cache.expiration: 5m

Configuration referenceedit

Specify the following options in the apm-server.rum section of the apm-server.yml config file:

enablededit

To enable RUM support, set apm-server.rum.enabled to true. By default this is disabled.

Enabling RUM support automatically enables Anonymous authentication. Anonymous access is required as the RUM agent runs in end users' browsers.

event_rate.limitedit

Deprecated in 7.15.0.

Replaced by <<config-auth-anon-event-limit

The maximum number of events allowed per second, per agent IP address.

Default: 300

event_rate.lru_sizeedit

Deprecated in 7.15.0.

Replaced by <<config-auth-anon-ip-limit

The number of unique IP addresses to track in an LRU cache. IP addresses in the cache will be rate limited according to the rate_limit.event_limit setting. Consider increasing this default if your site has many concurrent clients.

Default: 1000

allow_service_namesedit

Deprecated in 7.15.0.

Replaced by <<config-auth-anon-allow-service

A list of permitted service names for RUM support. Names in this list must match the agent’s service.name. This can be set to restrict RUM events to those with one of a set of known service names, in order to limit the number of service-specific indices or data streams created.

Default: Not set (any service name is accepted)

allow_originsedit

A list of permitted origins for RUM support. User-agents send an Origin header that will be validated against this list. This is done automatically by modern browsers as part of the CORS specification. An origin is made of a protocol scheme, host and port, without the URL path.

Default: ['*'] (allows everything)

allow_headersedit

HTTP requests made from the RUM agent to the APM Server are limited in the HTTP headers they are allowed to have. If any other headers are added, the request will be rejected by the browser due to Cross-Origin Resource Sharing (CORS) restrictions. Use this setting to allow additional headers. The default list of allowed headers includes "Content-Type", "Content-Encoding", and "Accept"; custom values configured here are appended to the default list and used as the value for the Access-Control-Allow-Headers header.

Default: []

response_headersedit

Custom HTTP headers to add to RUM responses. This can be useful for security policy compliance.

Values set for the same key will be concatenated.

Default: Not set

library_patternedit

RegExp to be matched against a stack trace frame’s file_name and abs_path attributes. If the RegExp matches, the stack trace frame is considered to be a library frame. When source mapping is applied, the error.culprit is set to reflect the function and the filename of the first non library frame. This aims to provide an entry point for identifying issues.

Default: "node_modules|bower_components|~"

exclude_from_groupingedit

RegExp to be matched against a stack trace frame’s file_name. If the RegExp matches, the stack trace frame is excluded from being used for calculating error groups.

Default: "^/webpack" (excludes stack trace frames that have a filename starting with /webpack)

source_mapping.enablededit

Used to enable/disable source mapping for RUM events. When enabled, the APM Server needs additional privileges to read source maps. See Create a writer user for more details.

Default: true

source_mapping.elasticsearchedit

Configure the Elasticsearch source map retrieval location, taking the same options as output.elasticsearch. This must be set when using an output other than Elasticsearch, and that output is writing to Elasticsearch. Otherwise leave this section empty.

source_mapping.cache.expirationedit

If a source map has been uploaded to the APM Server, source mapping is automatically applied to documents sent to the RUM endpoint. Source maps are fetched from Elasticsearch and then kept in an in-memory cache for the configured time. Values configured without a time unit are treated as seconds.

Default: 5m (5 minutes)

source_mapping.index_patternedit

Previous versions of APM Server stored source maps in apm-%{[observer.version]}-sourcemap indices. Search source maps stored in an older version with this setting.

Default: "apm-*-sourcemap*"

Ingest pipelinesedit

The default APM Server pipeline includes processors that enrich RUM data prior to indexing in Elasticsearch. See Parse data using ingest pipelines for details on how to locate, edit, or disable this preprocessing.