Load infoschema operation in domain takes a long time
Open
sig/sql-infra
type/enhancement
- Dominant language
- Go
- Stars
- 40.5k
- Forks
- 6.2k
- PR merge metrics
- PR metrics pending
Description
## Enhancement
Currently, when the load schema operation is executed on a cluster with hundreds of thousands of tables, the time and memory consumption are large. In particular, taking a long time can lead to some problems: https://github.com/pingcap/tidb/issues/45176
I tested the scenario myself:
```
Benchbot cluster, 1 TiDB, 3 TiKV, upgrade from v6.5.2 to v7.2.0
Make the first reload take 54s(create 500,000 tables ) in domain.Init for v7.2.0 branch.
```
At present, in addition to TiDB startup, the system information will be fully loaded, most of the other cases are incremental load table structure. But the problem still exists.
Contributor guide
Assessment
This issue has not been assessed yet.