Key operational procedures: Rolling restart (upgrade brokers one at a time, wait for ISR to recover between each). Preferred leader election (run after maintenance). Partition reassignment (kafka-reassign-partitions.sh for load balancing). Log retention tuning (adjust retention when disk is high). Consumer group offset management (reset with kafka-consumer-groups.sh --reset-offsets). Graceful shutdown (SIGTERM → broker completes in-flight requests before stopping).
Key configuration
# Graceful shutdown:
controlled.shutdown.enable=true
controlled.shutdown.max.retries=3