Goal:
- Possibility to run asyncio coroutines.
- Correct celery behavior on exceptions and task retries.
- Possibility to use aioredis lock.
So, how to run async tasks properly to achieve the goal?
What is RuntimeError: await wasn't used with future
(below), how can I fix it?
I have already tried:
1. asgiref
async_to_sync
(from asgiref https://pypi.org/project/asgiref/).
This option makes it possible to run asyncio coroutines, but retries functionality doesn’t work.
2. celery-pool-asyncio
(https://pypi.org/project/celery-pool-asyncio/)
Same problem as in asgiref. (This option makes it possible to run asyncio coroutines, but retries functionality doesn’t work.)
3. write own async to sync decorator
I have performed try to create my own decorator like async_to_sync that runs coroutines threadsafe (asyncio.run_coroutine_threadsafe
), but I have behavior as I described above.
4. asyncio module
Also I have try asyncio.run()
or asyncio.get_event_loop().run_until_complete()
(and self.retry(...)
) inside celery task. This works well, tasks runs, retries works, but there is incorrect coroutine execution – inside async
function I cannot use aioredis.
Implementation notes:
- start celery command:
celery -A celery_test.celery_app worker -l info -n worker1 -P gevent --concurrency=10 --without-gossip --without-mingle
- celery app:
transport = f"redis://localhost/9"
celery_app = Celery("worker", broker=transport, backend=transport,
include=['tasks'])
celery_app.conf.broker_transport_options = {
'visibility_timeout': 60 * 60 * 24,
'fanout_prefix': True,
'fanout_patterns': True
}
- utils:
@contextmanager
def temp_asyncio_loop():
# asyncio.get_event_loop() automatically creates event loop only for main thread
try:
prev_loop = asyncio.get_event_loop()
except RuntimeError:
prev_loop = None
loop = asyncio.new_event_loop()
asyncio.set_event_loop(loop)
try:
yield loop
finally:
loop.stop()
loop.close()
del loop
asyncio.set_event_loop(prev_loop)
def with_temp_asyncio_loop(f):
@functools.wraps(f)
def wrapper(*args, **kwargs):
with temp_asyncio_loop() as t_loop:
return f(*args, loop=t_loop, **kwargs)
return wrapper
def await_(coro):
return asyncio.get_event_loop().run_until_complete(coro)
- tasks:
@celery_app.task(bind=True, max_retries=30, default_retry_delay=0)
@with_temp_asyncio_loop
def debug(self, **kwargs):
try:
await_(debug_async())
except Exception as exc:
self.retry(exc=exc)
async def debug_async():
async with RedisLock(f'redis_lock_{datetime.now()}'):
pass
- redis lock
class RedisLockException(Exception):
pass
class RedisLock(AsyncContextManager):
"""
Redis Lock class
:param lock_id: string (unique key)
:param value: dummy value
:param expire: int (time in seconds that key will storing)
:param expire_on_delete: int (time in seconds, set pause before deleting)
Usage:
try:
with RedisLock('123_lock', 5 * 60):
# do something
except RedisLockException:
"""
def __init__(self, lock_id: str, value='1', expire: int = 4, expire_on_delete: int = None):
self.lock_id = lock_id
self.expire = expire
self.value = value
self.expire_on_delete = expire_on_delete
async def acquire_lock(self):
return await redis.setnx(self.lock_id, self.value)
async def release_lock(self):
if self.expire_on_delete is None:
return await redis.delete(self.lock_id)
else:
await redis.expire(self.lock_id, self.expire_on_delete)
async def __aenter__(self, *args, **kwargs):
if not await self.acquire_lock():
raise RedisLockException({
'redis_lock': 'The process: {} still run, try again later'.format(await redis.get(self.lock_id))
})
await redis.expire(self.lock_id, self.expire)
async def __aexit__(self, exc_type, exc_value, traceback):
await self.release_lock()
On my windows machine await redis.setnx(...)
blocks celery worker and it stops producing logs and Ctrl+C
doesn’t work.
Inside the docker container, I receive an error. There is part of traceback:
Traceback (most recent call last):
File "/usr/local/lib/python3.9/site-packages/aioredis/connection.py", line 854, in read_response
response = await self._parser.read_response()
File "/usr/local/lib/python3.9/site-packages/aioredis/connection.py", line 366, in read_response
raise ConnectionError(SERVER_CLOSED_CONNECTION_ERROR)
aioredis.exceptions.ConnectionError: Connection closed by server.
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/usr/local/lib/python3.9/site-packages/celery/app/trace.py", line 451, in trace_task
R = retval = fun(*args, **kwargs)
File "/usr/local/lib/python3.9/site-packages/celery/app/trace.py", line 734, in __protected_call__
return self.run(*args, **kwargs)
File "/usr/local/lib/python3.9/site-packages/celery/app/autoretry.py", line 54, in run
ret = task.retry(exc=exc, **retry_kwargs)
File "/usr/local/lib/python3.9/site-packages/celery/app/task.py", line 717, in retry
raise_with_context(exc)
File "/usr/local/lib/python3.9/site-packages/celery/app/autoretry.py", line 34, in run
return task._orig_run(*args, **kwargs)
File "/app/celery_tasks/tasks.py", line 69, in wrapper
return f(*args, **kwargs) # <--- inside with_temp_asyncio_loop from utils
...
File "/usr/local/lib/python3.9/contextlib.py", line 575, in enter_async_context
result = await _cm_type.__aenter__(cm)
File "/app/db/redis.py", line 50, in __aenter__
if not await self.acquire_lock():
File "/app/db/redis.py", line 41, in acquire_lock
return await redis.setnx(self.lock_id, self.value)
File "/usr/local/lib/python3.9/site-packages/aioredis/client.py", line 1064, in execute_command
return await self.parse_response(conn, command_name, **options)
File "/usr/local/lib/python3.9/site-packages/aioredis/client.py", line 1080, in parse_response
response = await connection.read_response()
File "/usr/local/lib/python3.9/site-packages/aioredis/connection.py", line 859, in read_response
await self.disconnect()
File "/usr/local/lib/python3.9/site-packages/aioredis/connection.py", line 762, in disconnect
await self._writer.wait_closed()
File "/usr/local/lib/python3.9/asyncio/streams.py", line 359, in wait_closed
await self._protocol._get_close_waiter(self)
RuntimeError: await wasn't used with future
- library versions:
celery==5.2.1
aioredis==2.0.0
2
Answers
Maybe it helps. https://github.com/aio-libs/aioredis-py/issues/1273
The main point is:
Use
solo
pool, then create a decorator which run task functionasyncio.get_event_loop().run_until_complete(f(*args, **kwargs))
and make your task asynchronous