// the find
coleifer/huey
a little task queue for python
Huey is a Python task queue from the author of peewee and other well-crafted single-file libraries. It covers the standard territory — delayed tasks, cron-style periodic tasks, retries, result storage — without pulling in Celery's machinery. The target is small-to-medium Django/Flask shops that find Celery over-engineered for their workload.
The storage backend abstraction is genuinely good: the same API works across Redis, SQLite, and in-memory, which means you can run tests without spinning up Redis and switch backends later without rewriting tasks. The execution model choice (process/thread/greenlet) is explicit and meaningful, not hidden behind a config flag. Django integration has kept pace with the official django.tasks backend interface, so you're not locked into huey's abstractions. The test suite is thorough and doesn't require external services for the bulk of it.
The consumer is a single process that owns all workers — there's no built-in distributed worker pool across machines, so horizontal scaling means running multiple consumers pointed at the same queue and hoping your tasks are safe to run concurrently. Result storage is opt-in and not guaranteed durable on the default Redis config (volatile memory); it's easy to silently drop results if Redis restarts. Task routing (sending specific tasks to specific queues/workers) is more limited than Celery's — you can't easily pin a slow job class to a dedicated worker pool without running a separate consumer instance. The 6k star count relative to Celery's 25k means the ecosystem of third-party extensions and Stack Overflow answers is thin.