A task queue for Python 3.7+ based on Redis Streams with a Celery-like API.
| Note: This is an alpha release. The project is under development, breaking changes are likely. |
|---|
- Supports both sync (e.g. Django, Flask) and async (e.g. Starlette, FastAPI) code.
- Sane defaults: at least once processing semantics, tasks acknowledged on completion.
- Automatic retries with exponential backoff for fire-and-forget jobs.
- Clear task statuses available (e.g. sent, executing, success).
- Automatic task discovery (defaults to using
**/tasks.py). - Exceptionally small and understandable codebase.
pip install fennelRun Redis and then execute your code in tasks.py:
from fennel import App
app = App(name='myapp', redis_url='redis://127.0.0.1')
@app.task
def foo(n):
return n
# Enqueue a task to be executed in the background by a fennel worker process.
foo.delay(7)Meanwhile, run the worker:
$ fennel worker --app tasks:appFennel also supports an async API. If your code is running in an event loop (e.g. via Starlette or FastAPI), you will want to use the async interface instead:
from fennel import App
app = App(name='myapp', redis_url='redis://127.0.0.1', interface='async')
@app.task
async def bar(x):
return x
await bar.delay(5)If you need to ensure that all tasks for a given key are processed in-order, please see our sister project Runnel.