LvYanquan created FLINK-35442: --------------------------------- Summary: Kafka Pipeline Sink may distribute the same key to different partitions after modify parallelism Key: FLINK-35442 URL: https://issues.apache.org/jira/browse/FLINK-35442 Project: Flink Issue Type: Improvement Components: Flink CDC Affects Versions: cdc-3.2.0 Reporter: LvYanquan Fix For: cdc-3.2.0
Currently, Kafka Pipeline Sink partition events by parallelInstanceId % partitions.length,this is because the previous partition operator has already shuffled events by key. However, If the parallelism of the task is modified and restarted, data with the same key before and after the task may be in different partitions. -- This message was sent by Atlassian Jira (v8.20.10#820010)