Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.
Original topic: empty-region-count不合并
[Test Environment for TiDB] Testing
[TiDB Version] v7.1.0
[Reproduction Path] None
[Encountered Issue: Problem Phenomenon and Impact]
Single-node mixed deployment, the overall cluster response slows down, and it is found that empty-region-count, miss-region-count, and undersized-region-count remain high.
Attempted to follow [FAQ] 解决已经开启 region merge， empty-region-count 仍然很多问题 - TiDB 的问答社区 for merging but it had no effect.
Is the missing peer 466k frantically catching up on replicas?
How many nodes? How many replicas are configured?
Are you suddenly creating many tables?
How long has this situation been going on?
Indeed, the network traffic is several megabytes at a time. It’s constantly uploading and downloading.
Single node, single replica. It should be a single replica, I’ll confirm it again tomorrow.
I don’t know, in the testing environment, the cluster response has been particularly slow recently, and I just discovered it.
It’s possible. This is a test environment where scripts for initializing and creating databases and tables are frequently tested. After testing, the database is deleted and the tests are run again.
Should it come down after a while?
I checked the data for a month, and it has been rising continuously.
Old Chinese medicine practitioner…
Single node with 3 replicas
miss-peer is related to replica settings.
It feels like the empty region not merging is related to TiFlash. Try adjusting the replicas first and then check again.
Should the replica count be set to 1?
I’ll try it on Monday. Since today is Friday, I won’t touch it for now.
Do not deploy with a single replica, PD might not handle it well. After all, no one uses it like this in production. 
No problem, this is a development and testing environment, a single replica is acceptable.
|
username: Kongdom
|
username: 
|
username: 
|
username: 
|
username: 
|
username: 
|
username: