We’re seeing a strange error on our Kafka cluster. Users have reported their consumer offsets seem to periodically reset, which makes downstream processes handle a massive amount of duplicates.
Our Kafka cluster is currently running 0.10.1.1. Previously the cluster had been at 0.10.0.1. It was upgraded to 0.10.1.0, but we ran into multiple deadlocks so we downgraded to 0.10.0.1. We upgrade from 0.10.0.1 to 0.10.1.1. This is the error we see. It only comes up every few weeks and only on some topics. [2017-04-12 08:41:14,313] ERROR [ReplicaFetcherThread-1-8], Error for partition [email_priority,10] to broker 8:org.apache.kafka.common.errors.UnknownServerException: The server experienced an unexpected error when processing the request (kafka.server.ReplicaFetcherThread) Has anyone seen this behavior before? Is there any way to prevent this from happening? Thank you all! Lawrence Weikum | Sr. Software Engineer | Pandora 1426 Pearl Street, Suite 100, Boulder CO 80302 m 720.203.1578 | lwei...@pandora.com
