When I perform a controlled shutdown of a Kafka Broker in a cluster, I see this exception in the kafka_root.logs for a short while:
org.apache.kafka.common.errors.NotLeaderForPartitionException: This server is not the leader for that topic-partition
Is this expected behavior? I thought that when Kafka brokers are shutdown cleanly, the cluster would elect another broker for each partition and inform consumers about it? Is this not the case?