TiKV Apply Wait Duration is Relatively High

Translated
1

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

Original topic: tikv apply wait duration 比较高

| username: Raymond

May I ask?
The write duration latency of TiKV is a bit high. After investigation, the latency is mainly high during the apply wait duration stage, indicating that there is a delay when the raft log is waiting to be applied, while the apply log duration is not high. Can this phenomenon be alleviated by increasing the number of threads in the apply-pool-size?
Additionally, could write hotspots also potentially cause the apply wait duration to be relatively high?

2
| username: hey-hoho | Original post link

Check the monitoring TiKV Details → Thread CPU → Async apply CPU to see if it is particularly high. Under normal circumstances, it should not exceed 90% * raftstore.apply-pool-size. If it is too high, you can consider increasing the apply-pool-size, provided that there are enough CPUs.

Write hotspots can cause this phenomenon. The apply log duration occurs after the apply wait duration. Currently, the tasks are all stuck at apply wait, so the apply log appears normal.

You can refer to this chart:

image
image1380×676 424 KB

3
| username: hey-hoho | Original post link

Are you working overtime so late, or are you studying? :joy:

4
| username: Raymond | Original post link

Thanks, indeed we need to check the CPU usage of apply.

5
| username: system | Original post link

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