Sling logs through Logback, configurable per-bundle from the Felix Web Console or via org.apache.sling.commons.log configurations. OSGi components can additionally expose JMX MBeans for runtime state, and a Prometheus exporter can scrape those into time-series metrics for dashboards and alerting.

Good dashboards track request latency and error rate at the HTTP layer, JVM health (heap, GC pauses, thread count), and repository-level indicators (Oak cache hit rate, revision GC lag). Alert on trends and thresholds that predict user-visible pain, not on every noisy log line.