tidb should avoid sending mpp tasks to tiflash which writes very slowly if possible
Open
type/feature-request
- Dominant language
- Go
- Stars
- 40.5k
- Forks
- 6.2k
- PR merge metrics
- PR metrics pending
Description
## Feature Request
The disk of one tiflash may be extremely slow, particularly when used on cloud. The wait index duration of this tiflash will also be very long, leading to slow queries. We need to find a way to identify such tiflash and choose another tiflash to run the mpp tasks if possible.
Contributor guide
Assessment
This issue has not been assessed yet.