Skip to main content

Monitoring

uxopian-ai exposes Spring Boot Actuator endpoints for health checks and metrics export. Metrics are collected by Micrometer and can be exported directly to OpenSearch.

Figure: Monitoring integration: Actuator endpoints and Micrometer metrics export to OpenSearch.

Actuator endpoints​

By default, the following endpoints are enabled:

EndpointPathPurpose
health/actuator/healthService health (UP/DOWN)
info/actuator/infoBuild info
loggers/actuator/loggersRuntime log level management

All paths are relative to the service's context path. In a standard production deployment the full URL is:

https://your-domain.example.com/gui/gateway/uxopian-ai/actuator/health

The gateway exposes this path publicly (no authentication required). See the security section of the gateway config in Docker Compose and Kubernetes.

To enable additional endpoints (e.g. prometheus, metrics, env), add to metrics.yml:

management:
    endpoints:
        web:
            exposure:
                include: health,info,loggers,prometheus

Metrics export​

uxopian-ai ships with Micrometer and can export metrics directly to OpenSearch. This is configured in config/metrics.yml:

management:
    elastic:
        metrics:
            export:
                enabled: true
                host: http://${opensearch.host}:${opensearch.port}
                index: micrometer-metrics
                auto-create-index: true
    endpoints:
        web:
            exposure:
                include: health,info,loggers
    metrics:
        uxopian-ai:
            enable: true
        enable:
            application: false
            tomcat: false
            logback: false
            jvm: false
            system: false
            http: false
            process: false
            disk: false
            executor: false

This writes uxopian-ai-specific metrics to the micrometer-metrics index in the same OpenSearch instance used for the knowledge base. Metrics for JVM, HTTP, Tomcat, and system internals are disabled by default to keep the index clean. Enable them individually if needed:

management:
    metrics:
        enable:
            jvm: true
            http: true

Kubernetes probes​

Configure liveness and readiness probes on the ai-standalone container using the health endpoint:

livenessProbe:
    httpGet:
        path: /gui/gateway/uxopian-ai/actuator/health
        port: 8080
    initialDelaySeconds: 60
    periodSeconds: 30
    failureThreshold: 3

readinessProbe:
    httpGet:
        path: /gui/gateway/uxopian-ai/actuator/health
        port: 8080
    initialDelaySeconds: 30
    periodSeconds: 10
    failureThreshold: 3

Add these under spec.containers[0] in the deployment, or pass them through the Helm chart's values.yaml if the chart exposes probe overrides.

The gateway service uses the same health path on port 8085.

Log level management​

The loggers endpoint allows changing log levels at runtime without a restart:

# List all loggers
curl http://localhost:8080/gui/gateway/uxopian-ai/actuator/loggers

# Set a specific logger to DEBUG
curl -X POST http://localhost:8080/gui/gateway/uxopian-ai/actuator/loggers/com.uxopian.ai \
  -H 'Content-Type: application/json' \
  -d '{"configuredLevel":"DEBUG"}'

This endpoint requires the ADMIN role in production (enforced by the gateway security config).

Last updated on