Downgrade each Kafka node one at a time. You can stop each Kafka broker and upgrade the component without downtime if you have enough replication for your topic.

Shut down the current Kafka daemon, switch to the previous version, and start the daemon:

su - kafka -c "/usr/hdp/current/kafka-broker/bin/kafka stop"

hdp-select set kafka-broker 2.2.2.0-2041

su - kafka -c "/usr/hdp/current/kafka-broker/bin/kafka start"

To verify that the Kafka daemon rejoined the cluster, run the validation steps in Appendix A.

Continue with the next step.