[ https://issues.apache.org/jira/browse/KAFKA-14043?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17606402#comment-17606402 ]
doupengwei commented on KAFKA-14043: ------------------------------------ Exp: for pg connector will try to release slot and publication resources, but due to net work issue. it will failed and try to update task status to failed. So i think we can add a generation information for task status, so that after net work issue resolved. the old status info can be filtered by check generation. > The latest status of the task is overwritten by the old status > -------------------------------------------------------------- > > Key: KAFKA-14043 > URL: https://issues.apache.org/jira/browse/KAFKA-14043 > Project: Kafka > Issue Type: Bug > Components: KafkaConnect > Affects Versions: 2.4.0 > Environment: Centos 7.6 > kafka 2.4.0 > Reporter: doupengwei > Priority: Major > > Kafka version : 2.4.0 > connect.protocol : compatible > In kafka connect cluster, if one node faced net work issue, then will > caused worker try to stop connector and task which running on this node. and > due to net work issue. it will stop failed and throw exception, then worker > process will try to update task or connector status, producer will retry > indefinitely until successfully sent. but due to net work issue, the new > assignment have performed on the other connect node. after net work recover, > old task status will cover latest status. -- This message was sent by Atlassian Jira (v8.20.10#820010)