flink cdc 2.2.0 flink 1.13.6有发现数据丢失的情况 劳烦敢问一下有没有什么解决方案呢?
版权声明:本文内容由阿里云实名注册用户自发贡献,版权归原作者所有,阿里云开发者社区不拥有其著作权,亦不承担相应法律责任。具体规则请查看《阿里云开发者社区用户服务协议》和《阿里云开发者社区知识产权保护指引》。如果您发现本社区中有涉嫌抄袭的内容,填写侵权投诉表单进行举报,一经查实,本社区将立刻删除涉嫌侵权内容。
If you are using Flink-CDC 2.3 to synchronize data from a source table with a "datetime" column type to Apache Hive, and the column type is changed to "long" in Hive, there are a few solutions you can consider to resolve the issue: 1 Cast the data: In Hive, you can cast the "long" data back to the "datetime" type using the cast() function in a SELECT statement. For example: SELECT cast(column_name as timestamp) as column_name ... 2 Modify the Flink-CDC configuration: You can modify the Flink-CDC configuration to explicitly specify the target column type in Hive as "datetime". This can be done by setting the "target.column.types" property in the Flink-CDC configuration file. 3 Use a data conversion tool: You can use a data conversion tool, such as Apache NiFi, to convert the data from "long" to "datetime" as it is being loaded into Hive. It is recommended to test any changes to the configuration or data conversion process thoroughly before implementing in a production environment to ensure that the data is correctly synchronized between the source and target systems.——该回答整理自钉群“Flink CDC 社区”
实时计算Flink版是阿里云提供的全托管Serverless Flink云服务,基于 Apache Flink 构建的企业级、高性能实时大数据处理系统。提供全托管版 Flink 集群和引擎,提高作业开发运维效率。