Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.
Original topic: 决定 coprocessor cpu 的参数是什么
[TiDB Usage Environment] Production Environment / Test / PoC
Production
[TiDB Version]
5.1.1
[Encountered Problem]
May I ask, when the unified thread pool is disabled, i.e., configured as follows:
readpool.coprocessor.use-unified-pool: false
readpool.storage.use-unified-pool: false
What parameter determines the coprocessor CPU? Is it readpool.storage.normal-concurrency?
The coprocessor CPU mainly handles read requests.
[Reproduction Path] What operations were performed to encounter the problem
[Problem Phenomenon and Impact]
[Attachments]
Please provide the version information of each component, such as cdc/tikv, which can be obtained by executing cdc version/tikv-server --version.
Set readpool.coprocessor, refer to: TiKV 配置文件描述 | PingCAP 文档中心
I found two on the official website, both of which handle read requests. What is the difference between them? Which one should be set?
readpool.storage
normal-concurrency
The number of threads in the thread pool that handle normal priority read requests.
When 8 ≤ cpu num ≤ 16, the default value is cpu_num * 0.5; when cpu num is greater than 8, the default value is 4; when cpu num is greater than 16, the default value is 8.
Minimum value: 1
readpool.coprocessor
normal-concurrency
The number of threads in the thread pool that handle normal priority Coprocessor requests.
Default value: CPU * 0.8
Minimum value: 1
It can be understood as point queries mainly through the storage pool CPU thread pool;
It can be understood as operator queries for push-down computation through the coprocessor pool CPU thread pool.
Currently, after version v4, TiDB recommends using the unified pool thread pool, allowing TiKV to adjust the CPU usage of storage and coprocessor by itself. You can refer to this document: TiKV 线程池性能调优 | PingCAP 文档中心
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.