Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.
Original topic: raft log 延迟过高
[TiDB Usage Environment] Online
[TiDB Version] 5.7.25-TiDB-v5.1.0
[Encountered Issue]
[Reproduction Path]
[Issue Phenomenon and Impact]
According to the official documentation, the response time for writing to RocksDB, which stores Raft logs, is significantly high.
The cluster is virtualized on SSDs.
How can I reduce read and write latency?
[Attachment]
It is already a solid-state drive, but virtualization is done on the solid-state drive here. I am not sure if the impact is significant.
There will also be speed differences with SSDs. It depends on whether the production environment can accept it. If not, then you still need to consider upgrading the hardware.
Got it, I’ll test the IO tonight.
It seems there is a parameter to configure RocksDB for asynchronous disk writes. Adjust this parameter and use sysbench to compare the results.
This parameter was removed after version 5.x.
It’s a pity that this parameter was removed.
From the disk latency monitoring, many points fall around 20ms, indicating that the disk latency is slightly high.
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.
|
username: Kongdom
|
username: 
|
username: 
|
username: