Preferred Leader Imbalance Reduces Cluster Efficiency
infoWhen partitions do not use their preferred leader, cluster load becomes unbalanced, reducing throughput and increasing latency as some brokers handle disproportionate leadership.
Monitor kafka.partition.leader_is_preferred = 0 across multiple partitions. Calculate percentage of partitions not using preferred leader. Alert when > 20% of partitions are not using preferred leader.
1. Run preferred leader election: Use kafka-preferred-replica-election.sh to restore preferred leaders. 2. Enable auto leader rebalance: Set auto.leader.rebalance.enable=true (if not already). 3. Check why elections aren't happening: Review leader.imbalance.check.interval.seconds. 4. Verify all brokers healthy: Ensure all brokers can accept leadership. 5. Monitor leader distribution: Use kafka.broker.leader_count to verify balanced distribution after election.