Disk exceeds 90%

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

Original topic: 磁盘超过90%

| username: 等一分钟

In which situations must the TiKV node data disk of TiDB be expanded when it exceeds 90%?

| username: 等一分钟 | Original post link

Sometimes it times out 90%, which affects the use of the cluster. However, my node is currently around 3TB, with 90% still having 300GB. Do I still need to expand?

| username: DBAER | Original post link

It might be related to this parameter:

| username: linuxmysql | Original post link

When disk space exceeds 80%, it’s time to consider adding more space. At 90%, it’s like being roasted over a fire :sweat_smile:

| username: 等一分钟 | Original post link

There are still more than 300GB left.

| username: WalterWj | Original post link

Expand quickly.

| username: Fly-bird | Original post link

In any situation where it exceeds 90%, expansion is necessary. Don’t worry about the cost.

| username: 江湖故人 | Original post link

It mainly depends on the data growth rate. When requesting resources, it’s best to provide an expected time frame for 100% capacity.

| username: Miracle | Original post link

Generally, you should consider scaling up when it reaches 80%. If it’s a default deployment, it seems like 5% of the space is reserved. Deleting that file can free up some disk space.

| username: 哈喽沃德 | Original post link

  1. Performance Impact: When the data disk is nearly full, disk read/write performance may be affected, leading to a decline in TiKV node performance. This can severely impact the database’s response speed and stability.
  2. Data Migration: Data in the TiKV cluster requires frequent migration, scheduling, and compaction operations. When disk space is tight, these operations may be restricted, affecting the overall data processing capability.
  3. Fault Recovery: If the data disk is very close to saturation, data scheduling and balancing may become more difficult after a node failure, making it challenging to effectively migrate and replicate data.
  4. Future Growth Needs: If the data disk usage is already close to 90%, the trend of future data growth may soon lead to insufficient disk space. Therefore, considering expansion early is a wise choice.
| username: 等一分钟 | Original post link

Okay, thank you everyone.

| username: forever | Original post link

Overall, it is proportional. Smaller total amounts grow more slowly, while larger total amounts grow faster. 80% is a critical value, and sudden situations need to be considered.

| username: TiDBer_5Vo9nD1u | Original post link

Shouldn’t 80% be enough to raise attention?

| username: TIDB-Learner | Original post link

If it has been used normally for ten years and only then exceeds 90%, with a total capacity of 3TB and 300GB remaining, there is no need to rush to expand capacity within a month or two. What about a year later?

| username: YuchongXU | Original post link

You can also look at the remaining space and growth trends.

| username: 像风一样的男子 | Original post link

Only new operations personnel would ask such a question, right? Who would wait until their disk is 90% full without expanding it, waiting for a crash to take the blame?

| username: redgame | Original post link

Scaling. Not waiting.

| username: TiDBer_aaO4sU46 | Original post link

What are you waiting for? If something goes wrong, you’ll take the blame.

| username: kelvin | Original post link

Don’t wait, you’ll have to take the blame if something goes wrong.

| username: 等一分钟 | Original post link

There is a premise, 90%, and more than 300GB, otherwise who would ask such a brainless question?