Table of Contents

Query cache

The query cache memoises the doc-id bitset for a query within an IndexSearcher. Repeat searches for the same query then skip the matching pass entirely.

Enable

var config = new IndexSearcherConfig
{
    EnableQueryCache     = true,   // off by default
    QueryCacheMaxEntries = 1024,   // LRU cap
};

using var searcher = new IndexSearcher(dir, config);

When it helps

  • Repeated filter clauses (e.g., status:active) inside larger boolean queries.
  • Hot dashboards re-issuing the same queries.

When it hurts

  • Highly varied queries with low repetition: pure overhead.
  • Memory-constrained environments: each cached entry stores a per-segment bitset.

Cache hits and misses are surfaced via RecordCacheHit and RecordCacheMiss. Watch the CacheHitRate in MetricsSnapshot to decide whether to keep it on.

Lifetime

The cache lives on the IndexSearcher. A SearcherManager refresh creates a new searcher with a fresh cache.

See also

Edit this page