[Q&A Review] TiCDC Source Code Analysis #2 | Introduction to the TiKV CDC Module

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

Original topic: 【Q&A 回顾】TiCDC 源码解读#2 | TiKV CDC 模块介绍

| username: TiDB社区小助手

This article is the second issue of TiCDC source code interpretation - TiCDC Scheduler working principle analysis, including the on-site Q&A collation, video review, and sharing material download collection. If you have more questions about this sharing, you are welcome to leave a message under this post.

The following is the Q&A review of this issue’s TiKV CDC module introduction

Q: Does ticdc support writing kafka compression?
A: Yes, by adding compression in the sink-uri, the supported compression algorithms are gzip, lz4, zstd, and snappy. The documentation has been updated Add `compression` to TiCDC Kafka sink config by overvenus · Pull Request #12198 · pingcap/docs-cn · GitHub

Q: How to evaluate the quality of tags and tag systems? What are the evaluation indicators?
A: (Did not understand the question, the user who asked the question can express it more accurately)

Q: How to know the time of transaction capture? And does it completely include the entire transaction?
A: TiCDC uses the resolved ts generated by TiKV, grouping all transaction change events with commit ts less than resolved ts by start ts, each group is a complete transaction.

Q: Does cdc need to interact with multiple tikvs to determine transactions?
A: If the regions involved in the transaction are distributed on multiple TiKVs, then interaction with these TiKVs is required.

Announcement of live interaction award-winning users

Congratulations to the following users who participated in the interaction and won the prize~ You can get 300 TiDB community points

The following users can add WeChat ID: Oneandtwii before December 10, reply: Your video account/Bilibili username + “TiDB community nickname”, to redeem points

Second issue interaction award-winning users
No. Video account user Bilibili user
1 Lula Bili_21085053
2 Account not canceled Melancholy Bean Bun
3 Alan AnKh93
4 D.O.T.Y|Image 16202742244_bili
5 Feng albiao
6 Healthy and Happy Xiao Jin Meat Mountain Demon King
7 liu
8 dbs

If you have more questions about this sharing, feel free to continue asking in this post.