我的索引包含超过3000万个文档。每个文档都有一个parentId字段和createdOn字段。修改后的文档有一个ModifyOn字段。日期存储为epoch_millis。如果在日期范围内至少创建或修改了一个“子”文档，则需要每个parentId查找前10个最新文档。我尝试使用复合聚合，但是要花很长时间才能完成。

[2019.11.05-19.47.41:442][  0]LogStreaming: Error: Couldn't find file for package /Script/DataLogger requested by async loading code. NameToLoad: /Script/DataLogger
[2019.11.05-19.47.41:442][  0]LogStreaming: Error: Found 1 dependent packages...
[2019.11.05-19.47.41:442][  0]LogStreaming: Error:   /Game/Blueprints/Tech/CustomGameInstance
[2019.11.05-19.47.41:446][  0]LogScript: Warning: accessed None
CustomGameInstance_C /Engine/Transient.GameEngine_0:CustomGameInstance_C_0
Function /Game/Blueprints/Tech/CustomGameInstance.CustomGameInstance_C:SetPlayerData:0245
[2019.11.05-19.47.41:446][  0]LogScript: Warning: Script call stack:
Function /Game/Blueprints/Tech/CustomGameInstance.CustomGameInstance_C:ReceiveInit
Function /Game/Blueprints/Tech/CustomGameInstance.CustomGameInstance_C:ExecuteUbergraph_CustomGameInstance
Function /Game/Blueprints/Tech/CustomGameInstance.CustomGameInstance_C:SetPlayerData

有更好的方法吗？