What are the main JMX metrics to monitor for a healthy Kafka cluster?

Essential JMX metrics: kafka.server:UnderReplicatedPartitions (should be 0), kafka.controller:ActiveControllerCount (should be exactly 1), kafka.server:OfflinePartitionsCount (0), kafka.network:RequestHandlerAvgIdlePercent (>0.3), kafka.network:NetworkProcessorAvgIdlePercent (>0.3), kafka.server:BytesInPerSec/BytesOutPerSec (throughput), kafka.log:LogEndOffset per partition (for lag tracking). Consumer-side: records-lag-max per consumer group. Set up Prometheus JMX Exporter for continuous monitoring.

Expected key concepts: UnderReplicatedPartitions, ActiveControllerCount, OfflinePartitionsCount, RequestHandlerAvgIdlePercent, BytesInPerSec, records-lag-max, Prometheus, JMX Exporter