Skip to content

v1.0.32 Performance, Observability, and Conflict Responses

Choose a tag to compare

View all tags
@orneryd orneryd released this 25 Mar 02:33
· 538 commits to main since this release
v1.0.32
f7a92fc

Hey Guys! Releases should start to slow down now as I get more queries optimized. with CI/CD it makes it easy to deploy as often as I need to as I am testing and deploying this for my own purposes and for my employer.

If you have any slow queries that can be optimized, please open an issue! I can only cover things I know about!

This NornicDB release focuses on one outcome: lower and more predictable latency on real production query shapes, with better Neo4j API compatibility and clearer operations visibility.

What users will notice:

  • Faster hot-path Cypher execution on common read/write templates:
    • direct ID lookup paths
    • key-list lookup paths (IN $keys, dual-key lookup patterns)
    • optional-match expansion flows
    • top-N list queries (ORDER BY ... LIMIT ...)
    • batched cleanup/delete flows (WITH ... LIMIT ... DETACH DELETE)
  • Lower contention during write-heavy periods:
    • lock scope reduced in write paths
    • embedding queue trigger behavior debounced/throttled to reduce interference with foreground writes
  • Better protocol-level observability:
    • Bolt and gRPC access logging now align with HTTP-style visibility
  • Better transaction semantics for Neo4j-compatible clients:
    • conflict/deadlock style failures now surface as Neo4j transient transaction errors (retryable classification), instead of syntax errors
  • Better tuning guidance:
    • expanded hot-path query cookbook with generic, production-oriented query templates
    • improved PROFILE diagnostics to show index usage decisions and rejection risk signals

In short: this release improves p95/p99 behavior on real graph workloads, reduces avoidable write-path interference, and makes client retry/ops behavior more predictable.

Assets 6
Loading