For a table with billions of rows, using the TiKV client in Java to subscribe to CDC events from over 4000 Regions, why do the Regions often have periods with no event push?

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

Original topic: 十亿级别的表,使用tikv client java 订阅 4000多个Region的cdc事件,Region的推送时常出现无任何事件推送是为什么?

| username: 迷人的Ti

[TiDB Usage Environment] Production Environment
[TiDB Version] 5.4.3
[Reproduction Path] Using tikv client java to simulate flinkcdc to subscribe to Region’s cdc events


But the received cdc events are a bit strange. Sometimes it will normally output cdc events for a period of time, and then it seems like the gRpc is interrupted, with no data at all. Sometimes it seems like it is directly interrupted, with no output at all, and sometimes it only outputs a little, but my source table has a lot of cdc events because it is continuously being written to.
[Encountered Problem: Problem Phenomenon and Impact]
Can you explain why this phenomenon occurs? How should it be adjusted? Another machine with the same deployment does not seem to have this problem, and the data output is very normal.
[Resource Configuration] 8c32g single machine deployed with a 5.4.3 cluster, one pd, one tidb, and three tikv.

| username: 大飞哥online | Original post link

No changes?

| username: 迷人的Ti | Original post link

The subscribed region has constant data changes.

| username: 芮芮是产品 | Original post link

5.45 Don’t use the old version.

| username: Billmay表妹 | Original post link

Sorry, I can’t translate images. Please provide the text you need translated.

| username: system | Original post link

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.