Core configuration options

A boolean specifying if the agent should be recording or not. When recording, the agent instruments incoming HTTP requests, tracks errors and collects and sends metrics. When not recording, the agent works as a noop, not collecting data and not communicating with the APM sever, except for polling the central configuration endpoint. Note that trace context propagation, baggage and log correlation will also be disabled when recording is disabled. As this is a reversible switch, agent threads are not being killed when inactivated, but they will be mostly idle in this state, so the overhead should be negligible.

You can use this setting to dynamically disable Elastic APM at runtime.

Setting to false will completely disable the agent, including instrumentation and remote config polling. If you want to dynamically change the status of the agent, use recording instead.

A boolean specifying if the agent should instrument the application to collect traces for the app. When set to false, most built-in instrumentation plugins are disabled, which would minimize the effect on your application. However, the agent would still apply instrumentation related to manual tracing options and it would still collect and send metrics to APM Server.

This is used to keep all the errors and transactions of your service together and is the primary filter in the Elastic APM user interface.

Instead of configuring the service name manually, you can also choose to rely on the service name auto-detection mechanisms of the agent. If service_name is set explicitly, all auto-detection mechanisms are disabled.

This is how the service name auto-detection works:

• For standalone applications

  • The agent uses Implementation-Title in the META-INF/MANIFEST.MF file if the application is started via java -jar.
  • Falls back to the name of the main class or jar file.

• For applications that are deployed to a servlet container/application server, the agent auto-detects the name for each application.

  • For Spring-based applications, the agent uses the spring.application.name property, if set.
  • For servlet-based applications, falls back to the Implementation-Title in the META-INF/MANIFEST.MF file.
  • Falls back to the display-name of the web.xml, if available.
  • Falls back to the servlet context path the application is mapped to (unless mapped to the root context).

Generally, it is recommended to rely on the service name detection based on META-INF/MANIFEST.MF. Spring Boot automatically adds the relevant manifest entries. For other applications that are built with Maven, this is how you add the manifest entries:

<build>
    <plugins>
        <plugin>
            <!-- replace with 'maven-war-plugin' if you're building a war -->
            <artifactId>maven-jar-plugin</artifactId>
            <configuration>
                <archive>
                    <!-- Adds
                    Implementation-Title based on ${project.name} and
                    Implementation-Version based on ${project.version}
                    -->
                    <manifest>
                        <addDefaultImplementationEntries>true</addDefaultImplementationEntries>
                    </manifest>
                    <!-- To customize the Implementation-* entries, remove addDefaultImplementationEntries and add them manually
                    <manifestEntries>
                        <Implementation-Title>foo</Implementation-Title>
                        <Implementation-Version>4.2.0</Implementation-Version>
                    </manifestEntries>
                    -->
                </archive>
            </configuration>
        </plugin>
    </plugins>
</build>

The service name must conform to this regular expression: ^[a-zA-Z0-9 _-]+$. In less regexy terms: Your service name must only contain characters from the ASCII alphabet, numbers, dashes, underscores and spaces.

If set, this name is used to distinguish between different nodes of a service, therefore it should be unique for each JVM within a service. If not set, data aggregations will be done based on a container ID (where valid) or on the reported hostname (automatically discovered or manually configured through hostname).

A version string for the currently deployed version of the service. If you don’t version your deployments, the recommended value for this field is the commit identifier of the deployed revision, e.g. the output of git rev-parse HEAD.

Similar to the auto-detection of service_name, the agent can auto-detect the service version based on the Implementation-Title attribute in META-INF/MANIFEST.MF. See service_name on how to set this attribute.

Allows for the reported hostname to be manually specified. If unset the hostname will be looked up.

The name of the environment this service is deployed in, e.g. "production" or "staging".

Environments allow you to easily filter data on a global level in the APM app. It’s important to be consistent when naming environments across agents. See environment selector in the APM app for more information.

By default, the agent will sample every transaction (e.g. request to your service). To reduce overhead and storage requirements, you can set the sample rate to a value between 0.0 and 1.0. (For pre-8.0 servers the agent still records and sends overall time and the result for unsampled transactions, but no context information, labels, or spans. When connecting to 8.0+ servers, the unsampled requests are not sent at all).

Value will be rounded with 4 significant digits, as an example, value 0.55555 will be rounded to 0.5556

Limits the amount of spans that are recorded per transaction.

This is helpful in cases where a transaction creates a very high amount of spans (e.g. thousands of SQL queries).

Setting an upper limit will prevent overloading the agent and the APM server with too much work for such edge cases.

A message will be logged when the max number of spans has been exceeded but only at a rate of once every 5 minutes to ensure performance is not impacted.

The following transaction, span, and error fields will be truncated at this number of unicode characters before being sent to APM server:

• transaction.context.request.body, error.context.request.body
• transaction.context.message.body, error.context.message.body
• span.context.db.statement

Note that tracing data is limited at the upstream APM server to max_event_size, which defaults to 300kB. If you configure long_field_max_length too large, it could result in transactions, spans, or errors that are rejected by APM server.

Sometimes it is necessary to sanitize the data sent to Elastic APM, e.g. remove sensitive data.

Configure a list of wildcard patterns of field names which should be sanitized. These apply for example to HTTP headers and application/x-www-form-urlencoded data.

This option supports the wildcard *, which matches zero or more characters. Examples: /foo/*/bar/*/baz*, *foo*. Matching is case insensitive by default. Prepending an element with (?-i) makes the matching case sensitive.

A list of instrumentations which should be selectively enabled. Valid options are annotations, annotations-capture-span, annotations-capture-transaction, annotations-traced, apache-commons-exec, apache-httpclient, asynchttpclient, aws-lambda, aws-sdk, cassandra, concurrent, dubbo, elasticsearch-restclient, exception-handler, executor, executor-collection, experimental, finagle-httpclient, fork-join, grails, grpc, hibernate-search, http-client, jakarta-websocket, java-ldap, javalin, javax-websocket, jax-rs, jax-ws, jdbc, jdk-httpclient, jdk-httpserver, jedis, jms, jsf, kafka, lettuce, log-correlation, log-error, log-reformatting, logging, micrometer, mongodb, mongodb-client, okhttp, opentelemetry, opentelemetry-annotations, opentelemetry-metrics, opentracing, process, public-api, quartz, rabbitmq, reactor, redis, redisson, render, scala-future, scheduled, servlet-api, servlet-api-async, servlet-api-dispatch, servlet-input-stream, servlet-service-name, servlet-version, sparkjava, spring-amqp, spring-mvc, spring-resttemplate, spring-service-name, spring-view-render, spring-webclient, spring-webflux, ssl-context, struts, timer-task, urlconnection, vertx, vertx-web, vertx-webclient, websocket. When set to non-empty value, only listed instrumentations will be enabled if they are not disabled through disable_instrumentations ([1.0.0]) or enable_experimental_instrumentations ([1.25.0]). When not set or empty (default), all instrumentations enabled by default will be enabled unless they are disabled through disable_instrumentations ([1.0.0]) or enable_experimental_instrumentations ([1.25.0]).

A list of instrumentations which should be disabled. Valid options are annotations, annotations-capture-span, annotations-capture-transaction, annotations-traced, apache-commons-exec, apache-httpclient, asynchttpclient, aws-lambda, aws-sdk, cassandra, concurrent, dubbo, elasticsearch-restclient, exception-handler, executor, executor-collection, experimental, finagle-httpclient, fork-join, grails, grpc, hibernate-search, http-client, jakarta-websocket, java-ldap, javalin, javax-websocket, jax-rs, jax-ws, jdbc, jdk-httpclient, jdk-httpserver, jedis, jms, jsf, kafka, lettuce, log-correlation, log-error, log-reformatting, logging, micrometer, mongodb, mongodb-client, okhttp, opentelemetry, opentelemetry-annotations, opentelemetry-metrics, opentracing, process, public-api, quartz, rabbitmq, reactor, redis, redisson, render, scala-future, scheduled, servlet-api, servlet-api-async, servlet-api-dispatch, servlet-input-stream, servlet-service-name, servlet-version, sparkjava, spring-amqp, spring-mvc, spring-resttemplate, spring-service-name, spring-view-render, spring-webclient, spring-webflux, ssl-context, struts, timer-task, urlconnection, vertx, vertx-web, vertx-webclient, websocket. For version 1.25.0 and later, use enable_experimental_instrumentations ([1.25.0]) to enable experimental instrumentations.

Whether to apply experimental instrumentations.

When reporting exceptions, un-nests the exceptions matching the wildcard pattern. This can come in handy for Spring’s org.springframework.web.util.NestedServletException, for example.

This option supports the wildcard *, which matches zero or more characters. Examples: /foo/*/bar/*/baz*, *foo*. Matching is case insensitive by default. Prepending an element with (?-i) makes the matching case sensitive.

A list of exceptions that should be ignored and not reported as errors. This allows to ignore exceptions thrown in regular control flow that are not actual errors

This option supports the wildcard *, which matches zero or more characters. Examples: /foo/*/bar/*/baz*, *foo*. Matching is case insensitive by default. Prepending an element with (?-i) makes the matching case sensitive.

Examples:

• com.mycompany.ExceptionToIgnore: using fully qualified name
• *ExceptionToIgnore: using wildcard to avoid package name
• *exceptiontoignore: case-insensitive by default

For transactions that are HTTP requests, the Java agent can optionally capture the request body (e.g. POST variables). For transactions that are initiated by receiving a message from a message broker, the agent can capture the textual message body.

If the HTTP request or the message has a body and this setting is disabled, the body will be shown as [REDACTED].

This option is case-insensitive.

Valid options: off, errors, transactions, all

If set to true, the agent will capture HTTP request and response headers (including cookies), as well as messages' headers/properties when using messaging frameworks like Kafka or JMS.

Labels added to all events, with the format key=value[,key=value[,...]]. Any labels set by application via the API will override global labels with the same keys.

A boolean specifying if the agent should instrument pre-Java-1.4 bytecode.

When set to true, disables log sending, metrics and trace collection. Trace context propagation and log correlation will stay active. Note that in contrast to disable_send the agent will still connect to the APM-server for fetching configuration updates and health checks.

Use to exclude specific classes from being instrumented. In order to exclude entire packages, use wildcards, as in: com.project.exclude.* This option supports the wildcard *, which matches zero or more characters. Examples: /foo/*/bar/*/baz*, *foo*. Matching is case insensitive by default. Prepending an element with (?-i) makes the matching case sensitive.

A list of methods for which to create a transaction or span.

If you want to monitor a large number of methods, use profiling_inferred_spans_enabled instead.

This works by instrumenting each matching method to include code that creates a span for the method. While creating a span is quite cheap in terms of performance, instrumenting a whole code base or a method which is executed in a tight loop leads to significant overhead.

Using a pointcut-like syntax, you can match based on

• Method modifier (optional)
  Example: public, protected, private or *
• Package and class name (wildcards include sub-packages)
  Example: org.example.*
• Method name (optional since 1.4.0)
  Example: myMeth*d
• Method argument types (optional)
  Example: (*lang.String, int[])
• Classes with a specific annotation (optional)
  Example: @*ApplicationScoped
• Classes with a specific annotation that is itself annotated with the given meta-annotation (optional)
  Example: @@javax.enterpr*se.context.NormalScope

The syntax is modifier @fully.qualified.AnnotationName fully.qualified.ClassName#methodName(fully.qualified.ParameterType).

A few examples:

• org.example.* [1.4.0]
• org.example.*#* (before 1.4.0, you need to specify a method matcher)
• org.example.MyClass#myMethod
• org.example.MyClass#myMethod()
• org.example.MyClass#myMethod(java.lang.String)
• org.example.MyClass#myMe*od(java.lang.String, int)
• private org.example.MyClass#myMe*od(java.lang.String, *)
• * org.example.MyClas*#myMe*od(*.String, int[])
• public org.example.services.*Service#*
• public @java.inject.ApplicationScoped org.example.*
• public @java.inject.* org.example.*
• public @@javax.enterprise.context.NormalScope org.example.*, public @@jakarta.enterprise.context.NormalScope org.example.*

Common configurations:

Trace all public methods in CDI-Annotated beans:

public @@javax.enterprise.context.NormalScope your.application.package.*
public @@jakarta.enterprise.context.NormalScope your.application.package.*
public @@javax.inject.Scope your.application.package.*

If trace_methods config option is set, provides a threshold to limit spans based on duration. When set to a value greater than 0, spans representing methods traced based on trace_methods will be discarded by default. Such methods will be traced and reported if one of the following applies:

• This method’s duration crossed the configured threshold.
• This method ended with Exception.
• A method executed as part of the execution of this method crossed the threshold or ended with Exception.
• A "forcibly-traced method" (e.g. DB queries, HTTP exits, custom) was executed during the execution of this method.

Set to 0 to disable.

Supports the duration suffixes ms, s and m. Example: 0ms.

When enabled, the agent will make periodic requests to the APM Server to fetch updated configuration. The frequency of the periodic request is driven by the Cache-Control header returned from APM Server/Integration, falling back to 5 minutes if not defined.

Disables the collection of breakdown metrics (span.self_time)

Sets the path of the agent config file. The special value _AGENT_HOME_ is a placeholder for the folder the elastic-apm-agent.jar is in. The file has to be on the file system. You can not refer to classpath locations.

A folder that contains external agent plugins.

Use the apm-agent-plugin-sdk and the apm-agent-api artifacts to create a jar and place it into the plugins folder. The agent will load all instrumentations that are declared in the META-INF/services/co.elastic.apm.agent.sdk.ElasticApmInstrumentation service descriptor. See integration-tests/external-plugin-test for an example plugin.

To enable distributed tracing, the agent adds trace context headers to outgoing requests (like HTTP requests, Kafka records, gRPC requests etc.). These headers (traceparent and tracestate) are defined in the W3C Trace Context specification.

When this setting is true, the agent will also add the header elastic-apm-traceparent for backwards compatibility with older versions of Elastic APM agents.

Use this option to disable tracecontext headers injection to any outgoing communication.

Sets the minimum duration of spans. Spans that execute faster than this threshold are attempted to be discarded.

The attempt fails if they lead up to a span that can’t be discarded. Spans that propagate the trace context to downstream services, such as outgoing HTTP requests, can’t be discarded. Additionally, spans that lead to an error or that may be a parent of an async operation can’t be discarded.

However, external calls that don’t propagate context, such as calls to a database, can be discarded using this threshold.

Supports the duration suffixes ms, s and m. Example: 0ms.

This config value allows you to specify which cloud provider should be assumed for metadata collection. By default, the agent will attempt to detect the cloud provider or, if that fails, will use trial and error to collect the metadata.

Valid options: AUTO, AWS, GCP, AZURE, NONE

A boolean specifying if the agent should search the class hierarchy for public api annotations (@CaptureTransaction, @CaptureSpan, @Traced and from 1.45.0 @WithSpan). When set to false, a method is instrumented if it is annotated with a public api annotation. When set to true methods overriding annotated methods will be instrumented as well. Either way, methods will only be instrumented if they are included in the configured application_packages.

With this option, you can group transaction names that contain dynamic parts with a wildcard expression. For example, the pattern GET /user/*/cart would consolidate transactions, such as GET /users/42/cart and GET /users/73/cart into a single transaction name GET /users/*/cart, hence reducing the transaction name cardinality.

This option supports the wildcard *, which matches zero or more characters. Examples: /foo/*/bar/*/baz*, *foo*. Matching is case insensitive by default. Prepending an element with (?-i) makes the matching case sensitive.

This option allows some control over how the APM agent handles W3C trace-context headers on incoming requests. By default, the traceparent and tracestate headers are used per W3C spec for distributed tracing. However, in certain cases it can be helpful to not use the incoming traceparent header. Some example use cases:

• An Elastic-monitored service is receiving requests with traceparent headers from unmonitored services.
• An Elastic-monitored service is publicly exposed, and does not want tracing data (trace-ids, sampling decisions) to possibly be spoofed by user requests.

Valid values are:

• continue: The default behavior. An incoming traceparent value is used to continue the trace and determine the sampling decision.
• restart: Always ignores the traceparent header of incoming requests. A new trace-id will be generated and the sampling decision will be made based on transaction_sample_rate. A span link will be made to the incoming traceparent.
• restart_external: If an incoming request includes the es vendor flag in tracestate, then any traceparent will be considered internal and will be handled as described for continue above. Otherwise, any traceparent is considered external and will be handled as described for restart above.

Starting with Elastic Observability 8.2, span links are visible in trace views.

This option is case-insensitive.

Valid options: continue, restart, restart_external

If any baggage key matches any of the patterns provided via this config option, the corresponding baggage key and value will be automatically stored on the corresponding transactions, spans and errors. The baggage keys will be prefixed with "baggage." on storage.