Hi, I am trying to setup a 3-node cluster using Kafka 3.0.0 and KRaft.
However, I always see the following error every millisecond in at least one of the bokers: 2021-10-06 08:24:49,309] INFO [ReplicaFetcher replicaId=3, leaderId=2, fetcherId=0] Node 2 was unable to process the fetch request with (sessionId=INVALID, epoch=INITIAL): INVALID_REQUEST. (org.apache.kafka.clients.FetchSessionHandler) The IDs vary between the error messages, but the general error is the same. I am using a PLAINTEXT listener for the communication between the nodes. All nodes are configured to be broker and controller: process.roles=broker,controller # We use the same id as already exists and we do always have controller+broker node.id=3 controller.quorum.voters=1@10.180.0.41:9095,2@10.180.0.42:9095,3@10.180.0.43:9095 Besides the error messages the communication seems to be working, partitions are distributed and in sync: Metadata for all topics (from broker 3: sasl_ssl://***********************:9090/3): 3 brokers: broker 1 at ******:9090 broker 2 at ******:9090 (controller) broker 3 at ******:9090 1 topics: topic "*********" with 8 partitions: partition 0, leader 2, replicas: 2,3, isrs: 2 partition 1, leader 3, replicas: 3,1, isrs: 3 partition 2, leader 1, replicas: 1,2, isrs: 1 partition 3, leader 2, replicas: 2,1, isrs: 2 partition 4, leader 1, replicas: 1,3, isrs: 1 partition 5, leader 3, replicas: 3,2, isrs: 3 partition 6, leader 3, replicas: 3,2, isrs: 3 partition 7, leader 2, replicas: 2,1, isrs: 2 However, when running kafkacat -L multiple times I can see that the controller is constantly switching between all brokers. What could be wrong?