Invalidating read cache cause high cpu usage #7259
Description
What type of bug is this?
Performance issue
What subsystems are affected?
Distributed Cluster, Standalone mode
Minimal reproduce step
This is observed from an environment that has around 100 regions and a large number of entries in the read cache.
The CPU usage is nearly 90% while executing a small number of queries.
What did you expect to see?
A lower CPU usage.
What did you see instead?
The cpu usage of datanode increased dramatically while executing queries. The flamegraph showed that the read cache is invalidating entries.
The flamegraph:
https://github.com/user-attachments/assets/6009346f-b67e-4d98-a58e-6b3e6bc3e3da
What operating system did you use?
Unrelated
What version of GreptimeDB did you use?
0.12