After adding a machine online, the disk usage is abnormal. Can we stop migrating regions?

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

Original topic: 线上挂了一台机器后,磁盘占用异常,可以停止迁移region吗

| username: zzw6776

【TiDB Usage Environment】Production Environment
Version 5.3
At 7:30 AM, one of the online machines suddenly went offline, and the disk usage increased abnormally.


According to the monitoring, it was due to region migration, which caused the disk usage to increase abnormally. In this situation, queries are particularly slow. Is there a way to stop the region migration first to meet business needs?

| username: Christophe | Original post link

You can scale down this machine, replace the disk, and then scale it back up.

| username: zzw6776 | Original post link

I feel that it’s not a problem of scaling in or out, but rather that the scheduling of regions is occupying the disk.

| username: ddhe9527 | Original post link

You can temporarily lower the region-schedule-limit and replica-schedule-limit, and then increase them during off-peak business hours to quickly replenish the replicas.

| username: zzw6776 | Original post link

In the group, an expert @李启航 provided a solution:
pd-ctl store limit all 0

| username: ddhe9527 | Original post link

666, limit from the consumer end: +1:

| username: tidb狂热爱好者 | Original post link

This method is good, it will not automatically schedule anymore.

| username: ddhe9527 | Original post link

Actually, I think this configuration, although effective, is a bit too extreme. What about newly created regions? Wouldn’t it be more appropriate to set enable-replace-offline-replica to false?

| username: 西伯利亚狼 | Original post link

I think it’s also this problem.

| username: zzw6776 | Original post link

Final conclusion:
The disk usage is all pulled up by TiFlash.
Therefore, adjusting enable-replace-offline-replica and enable-remove-down-replica is useless because these two parameters cannot restrict TiFlash.
Restricting the store can directly restrict TiFlash.

| username: ddhe9527 | Original post link

Is it the TiFlash node that is down?

| username: zzw6776 | Original post link

The machine that crashed has both TiKV and TiFlash.

| username: system | Original post link

This topic was automatically closed 1 minute after the last reply. No new replies are allowed.