[ https://issues.apache.org/jira/browse/KAFKA-14275?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17612452#comment-17612452 ]
Niket Goel commented on KAFKA-14275: ------------------------------------ PR – https://github.com/apache/kafka/pull/12709 > KRaft Controllers should crash after failing to apply any metadata record > -------------------------------------------------------------------------- > > Key: KAFKA-14275 > URL: https://issues.apache.org/jira/browse/KAFKA-14275 > Project: Kafka > Issue Type: Bug > Components: kraft > Affects Versions: 3.3.1 > Reporter: Niket Goel > Assignee: Niket Goel > Priority: Major > > When replaying records on a standby controller, any error encountered will > halt further processing of that batch. Currently we log an error and allow > the controller to continue normal operation. In contrast a similar error on > the active controller causes it to halt and exit the jvm. This is > inconsistent behavior as nothing prevents a standby from eventually becoming > the active controller (even when it had skipped over a record batch). We > should halt the process in the case of a standby controller as well. -- This message was sent by Atlassian Jira (v8.20.10#820010)