The incremental scan data of TiKV CDC will be affected by the most recent DDL. After a DDL occurs, only the incremental data after the DDL can be obtained, and the previous data will be lost

Translated
1

Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.

Original topic: Tikv CDC的增量扫数据会受到最近一次DDL的影响，发生DDL后，只能获取该DDL之后的增量数据，之前的数据将会丢失。

| username: 迷人的Ti

[Test Environment for TiDB] Testing
[TiDB Version] 5.4.3
[Reproduction Path] Use the TiKV client in Java to send a CDC request with the same startTs and receive a batch of incremental scan data. Then perform a DDL operation and send the request again with the same startTs, but only the incremental scan data after the DDL can be obtained.
[Encountered Problem: Phenomenon and Impact]
The incremental scan data of TiKV CDC is affected by the most recent DDL operation. After performing a DDL, only the incremental data after the DDL can be retrieved, and the data before the DDL will be lost.

2
| username: Fly-bird | Original post link

During the GC time, rerun the task based on TSO.

3
| username: 路在何chu | Original post link

Do you have operation logs? Please share them. I’ve been researching CDC recently.

4
| username: 迷人的Ti | Original post link

Can the TiKV client in Java set the GC time?

5
| username: 迷人的Ti | Original post link

Sorry, I didn’t keep it. The phenomenon has been the same in multiple tests.