Confluent - Administrator for Apache Kafka Study Guide

6: Observability

This chapter turns the broker fleet configured in the previous chapters into an observable system: it walks the JMX surface every broker exposes, names the specific MBeans that decide whether a cluster is healthy or degraded, measures consumer lag as the primary end-to-end signal, sizes storage through per-broker log directories, and covers the JVM, OS, and log-file signals that surround the JMX metrics. It closes with the alerting thresholds an administrator commits to on-call.