Skip to main content
Schema Registry Proxy exposes metrics in the Prometheus format that you can scrape and send to your external monitoring system.

​
Metrics endpoint

Schema Registry Proxy serves Prometheus metrics on a dedicated port:
cURL example
Set the port with OTEL_PROMETHEUS_PORT (default 9464).
Schema Registry Proxy only starts the metrics endpoint when OTEL_EXPORTER_OTLP_ENDPOINT is set. Without it, OpenTelemetry stays disabled and nothing listens on the Prometheus port, even though OTEL_PROMETHEUS_PORT has a default. Set OTEL_EXPORTER_OTLP_ENDPOINT to your collector address to enable the endpoint.
Setting OTEL_METRICS_ENABLED to false only stops Schema Registry Proxy from pushing metrics to the OTLP collector. The Prometheus endpoint stays available.

​
Request metrics

Schema Registry Proxy records these metrics for every request it handles. The histogram is exported as schema_registry_request_duration_seconds_bucket (with a le label), schema_registry_request_duration_seconds_sum and schema_registry_request_duration_seconds_count. The endpoint label holds the matched route pattern, for example /subjects/{subject}/versions, not the concrete URL. This keeps the number of label values small.

​
Authentication and license metrics

Alert on basic_auth_credentials_reload_total{result="failure"}. A failure means the credentials file on disk and the credentials in force have drifted apart. Alert on license_expiry_timestamp_seconds before a restart would fail. For example, this expression is true when the license expires in less than 30 days:

​
JVM runtime metrics

Schema Registry Proxy also exports the standard JVM runtime metrics from OpenTelemetry (memory, garbage collection, threads, CPU and class loading). For the list of metrics and their names, see the OpenTelemetry JVM metrics documentation.