I have a topic because of the wrong allocation of the number of shards when it was created. Use the delete command to delete it. Then, according to the method on the network, delete the corresponding node in / brokers/topics in the zk. You can"t see the topic through the list command. I think he should delete it.
then there were two broker that needed to be restarted yesterday. After the restart, it was found that the console output a large number of log contents as follows:
[2018-11-21 18:15:29,456] WARN [Replica Manager on Broker 103]: While recording the follower position, the partition [test-log,1] hasn"t been created, skip updating leader HW (kafka.server.ReplicaManager)
[2018-11-21 18:15:29,457] WARN [Replica Manager on Broker 103]: Fetch request with correlation id 1517033 from client ReplicaFetcherThread-0-103 on partition [test-log,1] failed due to Partition [test-log,1] doesn"t exist on 103 (kafka.server.ReplicaManager)
[2018-11-21 18:15:29,457] WARN [Replica Manager on Broker 103]: While recording the follower position, the partition [test-log,1] hasn"t been created, skip updating leader HW (kafka.server.ReplicaManager)
[2018-11-21 18:15:29,457] WARN [Replica Manager on Broker 103]: Fetch request with correlation id 1517034 from client ReplicaFetcherThread-0-103 on partition [test-log,1] failed due to Partition [test-log,1] doesn"t exist on 103 (kafka.server.ReplicaManager)
but at this time the topic can no longer be seen through list, and I have cleaned up the directory, but it will be printed out when I start it, which is very strange.
is there anyone who knows this problem