You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jira@kafka.apache.org by "Proven Provenzano (Jira)" <ji...@apache.org> on 2023/04/27 15:51:00 UTC

[jira] [Created] (KAFKA-14948) Broker fails to rejoin cluster when cluster is in dual write mode

Proven Provenzano created KAFKA-14948:
-----------------------------------------

             Summary: Broker fails to rejoin cluster when cluster is in dual write mode
                 Key: KAFKA-14948
                 URL: https://issues.apache.org/jira/browse/KAFKA-14948
             Project: Kafka
          Issue Type: Bug
          Components: kraft
    Affects Versions: 3.5.0
            Reporter: Proven Provenzano
            Assignee: Proven Provenzano


While testing migration dual write mode I came across this issue.

Initial setup: A single ZK node and a single Broker. Create a topic with some data.

Create a single controller and initiate migrations.

Update Broker to start migration.

Wait until all records are migrated. Cluster should be in dual write mode at this point.

Kill and restart the broker. Sometimes the broker will not rejoin the cluster and consuming records from a topic will fail to find topic.

Restarting the Controller will fix this issue.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)