You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "Yanquan Lv (Jira)" <ji...@apache.org> on 2025/09/18 12:17:00 UTC
[jira] [Updated] (FLINK-38380) MySQL cdc may skip data in the GTID disordered scenario even after GTID merged
[ https://issues.apache.org/jira/browse/FLINK-38380?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Yanquan Lv updated FLINK-38380:
-------------------------------
Description:
In [https://github.com/apache/flink-cdc/pull/3845,] we've merged restored gtid and server gtid to prevent data loss in most GTID misordering situations.
However, from the discuss of [https://github.com/apache/flink-cdc/pull/3845#discussion_r2298111455], the merge-based method does not address the risk of GTID loss at the beginning.
For example, if we have restored gtid of A:50-60 and server gtid of A:40-100, the merged result is A:40-60.
However, because of GTID disorder, it is uncertain whether the GTID interval A:40-49 was processed by the current read task.
was:
In [https://github.com/apache/flink-cdc/pull/3845,] we've merged restored gtid and server gtid to prevent data loss in most GTID misordering situations.
However, from the discuss of [https://github.com/apache/flink-cdc/pull/3845#discussion_r2298111455], the merge-based method does not address the risk of GTID loss at the beginning.
For example, if we have restored gtid of A:50-60 and server gtid of A:40-100, the merged result is A:40-60. However, because of GTID disorder, it is uncertain whether the GTID interval A:40-49 was processed by the current read task.
> MySQL cdc may skip data in the GTID disordered scenario even after GTID merged
> ------------------------------------------------------------------------------
>
> Key: FLINK-38380
> URL: https://issues.apache.org/jira/browse/FLINK-38380
> Project: Flink
> Issue Type: Improvement
> Components: Flink CDC
> Affects Versions: cdc-3.1.0, cdc-3.2.0, cdc-3.1.1, cdc-3.3.0, cdc-3.2.1, cdc-3.4.0
> Reporter: Yanquan Lv
> Priority: Major
>
> In [https://github.com/apache/flink-cdc/pull/3845,] we've merged restored gtid and server gtid to prevent data loss in most GTID misordering situations.
> However, from the discuss of [https://github.com/apache/flink-cdc/pull/3845#discussion_r2298111455], the merge-based method does not address the risk of GTID loss at the beginning.
> For example, if we have restored gtid of A:50-60 and server gtid of A:40-100, the merged result is A:40-60.
> However, because of GTID disorder, it is uncertain whether the GTID interval A:40-49 was processed by the current read task.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)