Package org.mockserver.metrics

Class OtelMetricsExporter

java.lang.Object

org.mockserver.metrics.OtelMetricsExporter

public class OtelMetricsExporter
extends Object

Optional exporter that publishes MockServer's explicitly-defined metrics (the same Metrics.Name gauges exposed for Prometheus) via OpenTelemetry OTLP. Off unless mockserver.otelMetricsEnabled is set.

Deliberately metrics-only: it registers one OTel observable gauge per Metrics.Name whose callback reads the current Metrics.get(org.mockserver.metrics.Metrics.Name) value at each collection. No spans, no auto-instrumentation. The OTel SDK is self-configured here (MockServer usually runs standalone), using the OTLP HTTP/protobuf exporter with the JDK HttpClient sender.

Method Summary

Modifier and Type	Method	Description
static OtelMetricsExporter	startIfEnabled()	Start the exporter if enabled in configuration, returning the running instance, or null if disabled or startup failed (fail-soft — telemetry must never prevent the server from running).
static OtelMetricsExporter	startIfEnabled(Configuration configuration)	As startIfEnabled() but preferring the values carried on the supplied Configuration instance, falling back to the static ConfigurationProperties store for each value the instance leaves unset.
static OtelMetricsExporter	startWithReader(io.opentelemetry.sdk.metrics.export.MetricReader reader)	Build a meter provider with the given reader and register observable instruments: One gauge per Metrics.Name (existing behaviour). JVM memory, thread, and GC gauges (same data as JvmMetricsCollector on the Prometheus side). A gauge mirroring the slow-request counter so OTLP-only consumers can observe it without a Prometheus scrape. A gauge mirroring the per-fault-type chaos-injection counter. A gauge mirroring the active service-scoped chaos count. A histogram for request-handling duration (seconds), fed from the same observation path as the Prometheus histogram. Visible for testing (a test can pass an in-memory reader instead of the OTLP periodic reader).
void	stop()	Stop exporting and release resources.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Details

startIfEnabled

public static OtelMetricsExporter startIfEnabled()

Start the exporter if enabled in configuration, returning the running instance, or null if disabled or startup failed (fail-soft — telemetry must never prevent the server from running).

startIfEnabled

public static OtelMetricsExporter startIfEnabled(Configuration configuration)

As startIfEnabled() but preferring the values carried on the supplied Configuration instance, falling back to the static ConfigurationProperties store for each value the instance leaves unset. This is what makes OTLP metric export settings applied programmatically or over PUT /mockserver/configuration actually take effect, rather than round-tripping through the DTO and being silently ignored here.

Parameters:

configuration - the live server configuration, or null to read the static store

startWithReader

public static OtelMetricsExporter startWithReader(io.opentelemetry.sdk.metrics.export.MetricReader reader)

Build a meter provider with the given reader and register observable instruments:

• One gauge per Metrics.Name (existing behaviour).
• JVM memory, thread, and GC gauges (same data as JvmMetricsCollector on the Prometheus side).
• A gauge mirroring the slow-request counter so OTLP-only consumers can observe it without a Prometheus scrape.
• A gauge mirroring the per-fault-type chaos-injection counter.
• A gauge mirroring the active service-scoped chaos count.
• A histogram for request-handling duration (seconds), fed from the same observation path as the Prometheus histogram.

Visible for testing (a test can pass an in-memory reader instead of the OTLP periodic reader).

stop

public void stop()

Stop exporting and release resources. Safe to call once.