Worker wakeups

When no work is available, a worker asks the queue backend to wait for a notification with WorkerConfig.poll_interval as the timeout:

from litestar_queues import QueueConfig, WorkerConfig

queue_config = QueueConfig(worker=WorkerConfig(poll_interval=0.25))

The effective wait ends when a backend notification arrives, the timeout expires, or worker shutdown is requested. Backends without notification support simply implement the timeout as polling.

Hints, not state

Redis and Valkey use pub/sub hints. SQLSpec turns on its best native wakeup transport automatically for any capable adapter (PostgreSQL LISTEN/NOTIFY for asyncpg, psycopg, and psqlpy; a durable in-process queue for DuckDB) and falls back to polling elsewhere. Advanced Alchemy can use PostgreSQL notification hints when enabled. The message only tells the worker to check. The saved task record has the real state.

Notifications may arrive late or not at all. The worker stays correct by checking the saved task state. Do not set the poll interval higher than the delay your service can tolerate.

Separate from task events

Worker wakeups are not QueueEvent delivery, browser fan-out, or durable event history. A Redis queue backend does not automatically configure Redis Channels for SSE or WebSocket consumers. See Task events.