The autocaching heuristics described here are not efficient for our use case. In our application, every search uses several terms queries that each return lists of size 10k-100k, which we want to be cached. We don't need more typical inverted index queries to be cached, as they are very fast as the documentation says.
With the default behavior, the cache provides very little benefit (an overall 5% hit rate). We turned on
index.queries.cache.everything and found that the cache hit rate is much better (around 50%) resulting in a 2x speed in average query time - but from the stats it seems like the cache churns constantly because too much is being cached. We'd much prefer to always cache these expensive terms queries while leaving the default behavior for everything else. Is there a way to do this? Thanks.