How to Locate the Issue of Individual Store Storage Capacity Being Overwhelmed

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

Original topic: 个别store存储容量占用极大的问题怎么定位

| username: 数据源的TiDB学习之路

[TiDB Usage Environment] Production Environment / Testing / PoC
[TiDB Version]
[Reproduction Path]
[Encountered Problem: Problem Phenomenon and Impact] In a certain customer’s environment, two store disks are full, while the utilization rate of other disks is poor. What are the troubleshooting ideas for this kind of problem?
[Resource Configuration] Go to TiDB Dashboard - Cluster Info - Hosts and take a screenshot of this page
[Attachment: Screenshot/Log/Monitoring]

| username: 扬仔_tidb | Original post link

It might be a region hotspot. Check the dashboard backend to see if there are any table hotspots.

| username: 有猫万事足 | Original post link

It might be a GC issue. Check out this post and see if it can resolve the problem.

| username: tidb菜鸟一只 | Original post link

When the usage rate of a normal node reaches around 80%, data will no longer be loaded onto this node and should be distributed to other nodes. Your usage rate is completely maxed out, which is not normal, right?

| username: xfworld | Original post link

Let’s take a look, what are the main issues with this scenario?

| username: Sword | Original post link

Check the number of regions in the store through PD and migrate some of them.

| username: zhaokede | Original post link

Take a look at the hotspots.

| username: 友利奈绪 | Original post link

Is it a hot topic?

| username: hacker_77powerful | Original post link

Which version of the database is it?

| username: 数据源的TiDB学习之路 | Original post link

Version 6.1

| username: yytest | Original post link

It should be a hot issue.

| username: zhaokede | Original post link

Region hotspot吧

| username: hacker_77powerful | Original post link

There should be a region hotspot. What problems might arise from using it this way?

| username: Jellybean | Original post link

It seems to be an issue with disk data imbalance.
The number of leaders and weights are balanced with other nodes, but the leader size and region size are significantly different. First, confirm the following:

  • Are there any differences in the hard drives of these nodes compared to others?
  • Confirm if the TiKV configurations of these nodes are different.
  • Are there any special PD scheduling policies specifically targeting these nodes?
| username: Jellybean | Original post link

Sorry, I can’t access external content such as the URL you provided. Please provide the text you need translated.

| username: kelvin | Original post link

Is this a hot topic?