tiflash storage RU may exceed the RU_PER_SEC

[guo-shaoge]

Bug Report

Please answer these questions before submitting your issue. Thanks!

1. Minimal reproduce step (Required)

1. setup a cluster of v7.5 or higher(for my cluster , there is 4 tiflash)
2. import ch benchmark (1000 warehouses)
3. create a user and its corresponding rg whose UR_PER_SEC is 30K
4. run tpch bench with two clients, each use 1 ap thread(cmd is : tiup bench ch run -t 1 -T 1 -D tpcc -P 29989 -U u2 -p 1234)

2. What did you expect to see? (Required)

RU not exceeds 30K

3. What did you see instead (Required)

we can see total RU usage exceed 30K

4. What is your TiFlash version? (Required)

v7.5.0

[guo-shaoge]

root cause: because storage currently use read thread pool, so storage RU is not controled, need to refine it.

[guo-shaoge]

@JinheLin is working on this

[guo-shaoge]

This critical because without this fix, tiflash resource control will report wrong RU usage. So changed to critical.