tool configuration

Engine-wide configuration: parallelism, caches, index storage, and any optional execution-mode switches.

Example

tool configuration {
  worker threads: 10
  use dedup cache: false
  index directory: "/etc/etl-engine.d/indexes/"
}

Properties

worker threads

Number of worker threads used to process source rows in parallel. Optional; the shipped configuration sets it to 10. Tune this to the source-data volume and the number of CPU cores available — see Performance Tuning.

index directory

Directory used to store and load index data and global sequences across runs. This is what makes it possible to split an ETL across multiple Rabbit-in-a-Hat files: the StoreIndexMap and StoreSequence table rules persist their state here.

use dedup cache

When true, caches deduplication results across rows. Optional; defaults to false. Enable this when many rows resolve to the same Normalize key — the cache turns a key re-check into a single hash lookup.

slot model

When "true", enables the slot-model executor. Optional; defaults to "false". The slot model is the recommended path for new deployments — it pre-resolves field-to-input mappings at compile time and removes several runtime locks. The legacy path remains for compatibility with rules that have not yet been migrated.

license file

Path to the engine's license file. Optional in production deployments — the shipped Docker image bakes in a default license-file location, so ordinary users do not set this key. Override it only if you are running the engine outside the shipped container or with a non-default license location.