The Apache Kafka monitoring and operations documentation highlights several controller-level metrics that are critical indicators of cluster health.
ActiveControllerCount must always be 1 in a healthy cluster. A value of 0 indicates no active controller, while values greater than 1 indicate a serious split-brain condition. This metric is fundamental for cluster stability.
OfflinePartitionsCount is one of the most important health metrics. Any value greater than 0 means that some partitions have no available leader, making them unavailable for producers and consumers. This directly indicates a degraded or unhealthy cluster.
UncleanLeaderElectionsPerSec tracks the rate of unclean leader elections. Unclean elections can cause data loss, and any non-zero value is a strong signal of broker failures or misconfiguration and should immediately trigger alerts.
The remaining options are not minimum health indicators. LastCommittedRecordOffset is not a standard health metric, and TopicsToDeleteCount only indicates pending administrative operations, not cluster instability.
Therefore, the three minimum and officially recommended broker metrics for monitoring Kafka cluster health are UncleanLeaderElectionsPerSec, ActiveControllerCount, and OfflinePartitionsCount.

max.poll.interval.ms defines themaximum delay between invocations of poll()before the consumer is considered failed. It essentially gives consumers more time to process records before needing to poll again.
FromKafka Consumer Configuration Reference:
"max.poll.interval.ms: The maximum delay between invocations of poll() when using consumer group management. If poll() is not called before expiration, the consumer is considered failed."
* session.timeout.ms and heartbeat.interval.ms relate to group coordination and heartbeats.
* fetch.max.wait.ms affects how long the broker waits to accumulate data before sending a fetch response.
Reference:Apache Kafka Consumer Configs

(Topic,Partition,Offset) uniquely identifies a message in Kafka

GlobalKTables have their datasets replicated on each Kafka Streams instance and therefore no repartitioning is required

Kafka clients useSLF4Jas a logging facade. To redirect Kafka logs tolog4j2, you must include theSLF4J binding for log4j2in your classpath:
"org.apache.logging.log4j:log4j-slf4j-impl bridges SLF4J to Log4j2 so that Kafka's logs integrate into your application's Log4j2 configuration."
* Option A is forLog4j 1.x, which is outdated.
* Option C is incorrect; Kafka clients don't include logging bindings.
* Option D is insufficient without the SLF4J bridge.
Reference:Kafka Logging Integration Practices