Examples

Every example here is a complete script that was run against a real Redis, and the output shown is what it printed. To run one yourself, change the Redis URL to match your setup and pip install rillkit (plus "rillkit[pydantic]" where noted).

Most scripts use start() and stop() with a short sleep so they finish by themselves. In a real service you would call broker.run() instead (see Lifecycle), and you would use the default block_ms and claim_idle_ms. The scripts use short values so the retry behaviour is visible in seconds.

Example	Shows
Basic worker	the smallest useful consumer, with the Message object
Two-stage pipeline	@subscribe plus @publisher: extraction, then feature engineering
Pydantic models	typed handlers and validation failures in the DLQ
Retries and DLQ	flaky and broken messages, dead letters, replay_dlq
Reject vs nack	nack() to retry, reject() to skip retries and dead-letter
Manual ack	AckPolicy.MANUAL with ack() and nack()
Timing and idempotency	a timing middleware and Idempotent
Multiple workers	several consumers sharing a group, active_workers
Graceful shutdown	SIGTERM, systemd and Docker
FastAPI publisher	publishing from a web endpoint
Migrating from redis-py	a hand-rolled XREADGROUP loop, before and after

The examples use stream names like orders, jobs and tasks. Use your own names, and a separate Redis database while experimenting.