@dbcube/core

Core runtime for the DBCube ecosystem. You normally don't install this package directly — it ships as a dependency of dbcube (the ORM), @dbcube/query-builder and @dbcube/schema-builder.

What it does

• Engine management: downloads, versions and runs the Rust query-engine and schema-engine binaries for your platform.
• Daemon mode: keeps the query engine alive as a local TCP server with warm connection pools (sub-millisecond dispatch), with automatic fallback to one-shot mode. Disable with DBCUBE_DAEMON=0.
• Config loading: parses dbcube.config.js (config.set({ databases: { ... } })) and exposes per-database settings.
• Pool tuning: the optional pool block per database — maxConnections, minConnections, acquireTimeoutMs, idleTimeoutMs and sessionIdleTimeoutMs.
• Processors: computed fields (@compute / @column in .cube files), trigger metadata and schema diffing used by table:refresh.

Resilient connections in containers

If Dbcube runs in a container that crashes and restarts repeatedly, connections left behind by each ungraceful death can pile up on the database until it runs out. pool.sessionIdleTimeoutMs (default 300000 ms = 5 min) tells the database itself to drop sessions left idle longer than that — mapping to wait_timeout (MySQL), idle_session_timeout (PostgreSQL 14+) and maxIdleTimeMS (MongoDB) — so orphans are reaped in minutes instead of hours. Live pooled connections are unaffected. Set it to 0 to disable. Requires query-engine v1.1.3+. Full details on dbcube.dev.

Usage

const { QueryEngine } = require("@dbcube/core");

const engine = new QueryEngine("myapp"); // name from dbcube.config.js
const res = await engine.rawQuery("SELECT 1");

For application code use the dbcube ORM instead.

Documentation

https://dbcube.dev

License

MIT