Under some workload the `multiget_data` maybe stuck and raise rpc timeout. Inorder to to minimize the latency of point queries, so it should be placed in `_light_work_pool`, since it's typically finished in hundreds of millisecond
Under some workload the `multiget_data` maybe stuck and raise rpc timeout. Inorder to to minimize the latency of point queries, so it should be placed in `_light_work_pool`, since it's typically finished in hundreds of millisecond