[Feature Request] Interface Fixes for Star Tree Search #15604
Description
Is your feature request related to a problem? Please describe
Currently, there are scenarios where the number of star tree documents can exceed the segment max doc. In such cases, the caching of the query is not possible because we create a FixedBitSet with segment max doc, it can overflow when we perform a set operation for documents more than the star tree docs.
Describe the solution you'd like
This change should introduce two key improvements to the search flows:
Caching Decision Enhancement: The search flows now have access to the StarTreeMetadata, allowing them to make an informed decision on whether to cache a query or not. This decision is based on the segment's maximum document count and the number of star tree documents.
Graceful Handling of Missing Dimension Values: Previously, an assertion was raised when a child node corresponding to a dimension value was not found. This change removes that assertion, enabling the system to gracefully handle scenarios where a dimension value does not exist in the children.
Related component
Search:Aggregations
Describe alternatives you've considered
No response
Additional context
No response