apache kafka - Leader election of partitions not happening automatically for one of the broker

Question

Our Kafka server setup is 3 zookeeper and 3 broker. For a topic with replication factor 1, all partitions(leaders) on a particular broker are going to -1 state.

We have enabled unclean.leader.election.enable=true and auto.leader.rebalance.enable=true and tried to increase replica and reassign the partition but not able to succeed.

FYI This all started when one of the broker keeps of restarting due to Out of Memory error and for now we increased the min memory allocation to overcome this(6GB to 8GB). Remaining two brokers are fine and leaders of those partition are fine.

We are still not able to figure out root cause for the leader state to -1 cause and memory issue.

Answer 1

等待大神答复