Does the empty-region-count parameter data under region health have any significant impact on the database?

Translated
1

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

Original topic: region health下的empty-region-count参数数据大有对数据库什么影响吗？

| username: TiDBer_Y2d2kiJh

[TiDB Usage Environment] Production Environment / Testing / PoC
[TiDB Version] v5.4.0 3tidb 3pd 3tikv
[Reproduction Path] The empty-region-count parameter under region health shows 165. What is the purpose of this parameter, and does it have any impact on the database?
[Encountered Problem: Problem Phenomenon and Impact]
[Resource Configuration] Enter TiDB Dashboard - Cluster Info - Hosts and take a screenshot of this page
[Attachment: Screenshot/Log/Monitoring]

regionhealth
regionhealth1380×200 23.7 KB

2
| username: Soysauce520 | Original post link

Excessive influence on PD scheduling, empty regions also need to send heartbeats. It is recommended to increase the PD parameter merge.

3
| username: tidb菜鸟一只 | Original post link

A large number of empty regions consume network, CPU, and memory resources, and also affect your cluster’s load balancing. Imagine you have 3 TiKV nodes, but most of the regions on one of the TiKV nodes are empty. This means that the load cannot be effectively distributed to this node, essentially leaving only 2 TiKV nodes working, which goes against the purpose of a distributed database.

However, since you only have a few hundred regions, the problem is not significant. You can wait for them to merge automatically, or you can adjust the region merge parameters to speed up the merging process, or enable cross-table merging.

4
| username: 有猫万事足 | Original post link

Did you use partitioned tables when creating the table? If so, having empty regions when no data is imported is normal. Most of the empty regions that cannot be merged by PD are due to this reason.