Consider a single-threaded Python program. A coroutine named "first" is blocked on I/O. The subsequent instruction is "await second." Is the coroutine "second" guaranteed to execute immediately until it blocks on I/O? Or, can "first" resume executing (due to the I/O operation completing) before "second" is invoked?
Asyncio implemented a way that second would start executing until it would return control to event loop (it usually happens when it reaches some I/O operation) and only after it first can be resumed. I don't think it somehow guaranteed to you, but hardly believe this implementation will be changed either.
If for some reason you don't want first to resume executing until some part of second reached, it's probably better explicitly to use Lock to block first from executing before moment you want.
Example to show when control returns to event loop and execution flow can be changed:
import asyncio
async def async_print(text):
print(text)
async def first():
await async_print('first 1')
await async_print('first 2')
await asyncio.sleep(0) # returning control to event loop
await async_print('first 3')
async def second():
await async_print('second 1')
await async_print('second 2')
await asyncio.sleep(0) # returning control to event loop
await async_print('second 3')
async def main():
asyncio.ensure_future(first())
asyncio.ensure_future(second())
await asyncio.sleep(1)
loop = asyncio.new_event_loop()
asyncio.set_event_loop(loop)
try:
loop.run_until_complete(main())
finally:
loop.run_until_complete(loop.shutdown_asyncgens())
loop.close()
Output:
first 1
first 2
second 1
second 2
first 3
second 3
Related
Can someone explain this behaviour (python 3.11):
import asyncio as aio
async def cancel_me():
while True:
try:
await aio.sleep(1) # 1. sleep
except aio.CancelledError:
print(f"Cancelled {aio.current_task().get_name()}")
try:
await aio.sleep(1) # 2. sleep
except aio.CancelledError:
print("Second sleep cancelled")
await aio.sleep(1) # 3. sleep
print("Third sleep NOT cancelled")
raise
async def main():
tsk = aio.Task(cancel_me())
await aio.sleep(0.1)
tsk.cancel()
aio.run(main())
Running it gives:
Cancelled Task-2
Second sleep cancelled
Third sleep NOT cancelled
Why is the second sleep (from the top) cancelled while the third one runs to completion?
I was expecting either that both would be cancelled or none,
updated
The behavior does seem to be an arbitrariness resulting from an implementation detail. The second, but not third sleep is cancelled, even if it is moved out of the except block. If a second sleep is called in main prior to that, however, the second sleep in cancel_me runs fine.
This is the snippet I've been playing around, commenting parts and changing counters between executions to check the various outcomes:
async def cancel_me():
for i in range(10):
try:
await aio.sleep(1)
except aio.CancelledError:
print(f"sleep canceled at counter = {i}")
else:
print(f"Ok at {i}")
async def main():
tsk = aio.Task(cancel_me())
for i in range(1):
await aio.sleep(0.1)
print(f"sending cancellation {i}")
tsk.cancel()
#if i == 0:
#await aio.sleep(4)
aio.run(main())
The workaround I ended up with was:
import asyncio as aio
async def cancel_me():
while True:
try:
await aio.sleep(0.1) # 1. sleep (do regular Task stuff)
except aio.CancelledError:
# Async cleanup
print(f"Cancelled {aio.current_task().get_name()}")
try:
await aio.sleep(0.1) # 2. sleep (ignore strange behaviour)
except aio.CancelledError:
pass
await aio.sleep(1) # 3. sleep (async cleanup)
return
async def main():
tsk = aio.Task(cancel_me())
await aio.sleep(0.1)
tsk.cancel()
aio.run(main())
This ensures that the strange behaviour is ignored at the cost of a 0.1 second delay, before commencing async cleanup. This is future proof in that once this quirk is fixed the code still runs.
Imagine that I have a main program which starts many async activities which all wait on queues to do jobs, and then on ctrl-C properly closes them all down: it might look something like this:
async def run_act1_forever():
# this is the async queue loop
while True:
job = await inputQueue1.get()
# do something with this incoming job
def run_activity_1(loop):
# run the async queue loop as a task
coro = loop.create_task(run_act1_forever())
return coro
def mainprogram():
loop = asyncio.get_event_loop()
act1 = run_activity_1(loop)
# also start act2, act3, etc here
try:
loop.run_forever()
except KeyboardInterrupt:
pass
finally:
act1.cancel()
# also act2.cancel(), act3.cancel(), etc
loop.close()
This all works fine. However, starting up activity 1 is actually more complex than this; it happens in three parts. Part 1 is to wait on the queue until a particular job comes in, one time; part 2 is a synchronous part which has to run in a thread with run_in_executor, one time, and then part 3 is the endless waiting on the queue for jobs as above. How do I structure this? My initial thought was:
async def run_act1_forever():
# this is the async queue loop
while True:
job = await inputQueue1.get()
# do something with this incoming job
async def run_act1_step1():
while True:
job = await inputQueue1.get()
# good, we have handled that first task; we're done
break
def run_act1_step2():
# note: this is sync, not async, so it's in a thread
# do whatever, here, and then exit when done
time.sleep(5)
def run_activity_1(loop):
# run step 1 as a task
step1 = loop.create_task(run_act1_step1())
# ERROR! See below
# now run the sync step 2 in a thread
self.loop.run_in_executor(None, run_act1_step2())
# finally, run the async queue loop as a task
coro = loop.create_task(run_act1_forever())
return coro
def mainprogram():
loop = asyncio.get_event_loop()
act1 = run_activity_1(loop)
# also start act2, act3, etc here
try:
loop.run_forever()
except KeyboardInterrupt:
pass
finally:
act1.cancel()
# also act2.cancel(), act3.cancel(), etc
loop.close()
but this does not work, because at the point where we say "ERROR!", we need to await the step1 task and we never do. We can't await it, because run_activity_1 is not an async function. So... what should I do here?
I thought about getting the Future back from calling run_act1_step1() and then using future.add_done_callback to handle running steps 2 and 3. However, if I do that, then run_activity_1() can't return the future generated by run_act1_forever(), which means that mainprogram() can't cancel that run_act1_forever() task.
I thought of generating an "empty" Future in run_activity_1() and returning that, and then making that empty Future "chain" to the Future returned by run_act1_forever(). But Python asyncio doesn't support chaining Futures.
You say that things are difficult because run_activity_1 is not an async function, but don't really detail why it can't be async.
async def run_activity_1(loop):
await run_act1_step1()
await loop.run_in_executor(None, run_act1_step2)
await run_act1_forever()
The returned coroutine won't be the same as the one returned by run_act1_forever(), but cancellation should propagate if you've got as far as executing that step.
With this change, run_activity_1 is no longer returning a task, so the invocation inside mainprogram would need to change to:
act1 = loop.create_task(run_activity_1(loop))
I think you were on the right track when you said, "I thought about getting the Future back from calling run_act1_step1() and then using future.add_done_callback to handle running steps 2 and 3." That's the logical way to structure this application. You have to manage the various returned objects correctly, but a small class solves this problem.
Here is a program similar to your second code snippet. It runs (tested with Python3.10) and handles Ctrl-C gracefully.
Python3.10 issues a deprecation warning when the function asyncio.get_event_loop() is called without a running loop, so I avoided doing that.
Activities.run() creates task1, then attaches a done_callback that starts task2 and the rest of the activities. The Activities object keeps track of task1 and task2 so they can be cancelled. The main program keeps a reference to Activities, and calls cancel_gracefully() to do the right thing, depending on how far the script progressed through the sequence of start-up activities.
Some care needs to be taken to catch the CancelledExceptions; otherwise stuff gets printed on the console when the program terminates.
The important difference between this program and your second code snippet is that this program immediately stores task1 and task2 in variables so they can be accessed later. Therefore they can be cancelled any time after their creation. The done_callback trick is used to launch all the steps in the proper order.
#! python3.10
import asyncio
import time
async def run_act1_forever():
# this is the async queue loop
while True:
await asyncio.sleep(1.0)
# job = await inputQueue1.get()
# do something with this incoming job
print("Act1 forever")
async def run_act1_step1():
while True:
await asyncio.sleep(1.0)
# job = await inputQueue1.get()
# good, we have handled that first task; we're done
break
print("act1 step1 finished")
def run_act1_step2():
# note: this is sync, not async, so it's in a thread
# do whatever, here, and then exit when done
time.sleep(5)
print("Step2 finished")
class Activities:
def __init__(self, loop):
self.loop = loop
self.task1: asyncio.Task = None
self.task2: asyncio.Task = None
def run(self):
# run step 1 as a task
self.task1 = self.loop.create_task(run_act1_step1())
self.task1.add_done_callback(self.run2)
# also start act2, act3, etc here
def run2(self, fut):
try:
if fut.exception() is not None: # do nothing if task1 failed
return
except asyncio.CancelledError: # or if it was cancelled
return
# now run the sync step 2 in a thread
self.loop.run_in_executor(None, run_act1_step2)
# finally, run the async queue loop as a task
self.task2 = self.loop.create_task(run_act1_forever())
async def cancel_gracefully(self):
if self.task2 is not None:
# in this case, task1 has already finished without error
self.task2.cancel()
try:
await self.task2
except asyncio.CancelledError:
pass
elif self.task1 is not None:
self.task1.cancel()
try:
await self.task1
except asyncio.CancelledError:
pass
# also act2.cancel(), act3.cancel(), etc
def mainprogram():
loop = asyncio.new_event_loop()
asyncio.set_event_loop(loop)
acts = Activities(loop)
loop.call_soon(acts.run)
try:
loop.run_forever()
except KeyboardInterrupt:
pass
loop.run_until_complete(acts.cancel_gracefully())
if __name__ == "__main__":
mainprogram()
You can do this with a combination of threading events and asyncio events. You'll need two events, one to signal the first item has arrived. The thread will wait on this event, so it needs to be a threading Event. You'll also need one to signal the thread is finished. Your run_act1_forever coroutine will await this, so it will need to be an asyncio Event. You can then return the task for run_act1_forever normally and cancel it as you need.
Note that when setting the asyncio event from the separate thread you'll need to use loop.call_soon_threadsafe as asyncio Events are not thread safe.
import asyncio
import time
import threading
import functools
from asyncio import Queue, AbstractEventLoop
async def run_act1_forever(inputQueue1: Queue,
thread_done_event: asyncio.Event):
await thread_done_event.wait()
print('running forever')
while True:
job = await inputQueue1.get()
async def run_act1_step1(inputQueue1: Queue,
first_item_event: threading.Event):
print('Waiting for queue item')
job = await inputQueue1.get()
print('Setting event')
first_item_event.set()
def run_act1_step2(loop: AbstractEventLoop,
first_item_event: threading.Event,
thread_done_event: asyncio.Event):
print('Waiting for event...')
first_item_event.wait()
print('Got event, processing...')
time.sleep(5)
loop.call_soon_threadsafe(thread_done_event.set)
def run_activity_1(loop):
inputQueue1 = asyncio.Queue(loop=loop)
first_item_event = threading.Event()
thread_done_event = asyncio.Event(loop=loop)
loop.create_task(run_act1_step1(inputQueue1, first_item_event))
inputQueue1.put_nowait('First item to test the code')
loop.run_in_executor(None, functools.partial(run_act1_step2,
loop,
first_item_event,
thread_done_event))
return loop.create_task(run_act1_forever(inputQueue1, thread_done_event))
def mainprogram():
loop = asyncio.new_event_loop()
act1 = run_activity_1(loop)
# also start act2, act3, etc here
try:
loop.run_forever()
except KeyboardInterrupt:
pass
finally:
act1.cancel()
# also act2.cancel(), act3.cancel(), etc
loop.close()
mainprogram()
Why FastAPI doesn't complain about the following task
app.state["my_task"] = asyncio.create_task(my_task())
not being awaited in any part of the code?
app is an instance of FastAPI() whereas app.state a simple dictionary.
I can even app.state["my_task"].cancel() in a shutdown callback.
From the asyncio docs:
When a coroutine function is called, but not awaited (e.g. coro() instead of await coro()) or the coroutine is not scheduled with asyncio.create_task(), asyncio will emit a RuntimeWarning
Thus, it is expected that, when using create_task(), no warning is emmited. This is no exclusive behavior to FastAPI. You can check it in the minimal snippet below:
import asyncio
async def task():
print("foobar")
async def main():
# task() # Emits a RuntimeWarning
# await task() # Does not emmit
x = asyncio.create_task(task()) # Doesn't either
loop = asyncio.get_event_loop()
loop.run_until_complete(main())
Edit PS: Calling the coroutine function as-is does not schedule the underlying task (and returns a coroutine object), while calling create_task() does (and returns a Task object). For more info, again, check the asyncio docs.
I'm trying to expose an event-based communication as a coroutine. Here is an example:
class Terminal:
async def start(self):
loop = asyncio.get_running_loop()
future = loop.create_future()
t = threading.Thread(target=self.run_cmd, args=future)
t.start()
return await future
def run_cmd(self, future):
time.sleep(3) # imitating doing something
future.set_result(1)
But when I run it like this:
async def main():
t = Terminal()
result = await t.start()
print(result)
asyncio.run(main())
I get the following error: RuntimeError: await wasn't used with future
Is it possible to achieve the desired behavior?
There are two issues with your code. One is that the args argument to the Thread constructor requires a sequence or iterable, so you need to write wrap the argument in a container, e.g. args=(future,). Since future is iterable (for technical reasons unrelated to this use case), args=future is not immediately rejected, but leads to the misleading error later down the line.
The other issue is that asyncio objects aren't thread-safe, so you cannot just call future.set_result from another thread. This causes the test program to hang even after fixing the first issue. The correct way to resolve the future from another thread is through the call_soon_threadsafe method on the event loop:
class Terminal:
async def start(self):
loop = asyncio.get_running_loop()
future = loop.create_future()
t = threading.Thread(target=self.run_cmd, args=(loop, future,))
t.start()
return await future
def run_cmd(self, loop, future):
time.sleep(3)
loop.call_soon_threadsafe(future.set_result, 1)
If your thread is really just calling a blocking function whose result you're interested in, consider using run_in_executor instead of manually spawning threads:
class Terminal:
async def start(self):
loop = asyncio.get_running_loop()
return await loop.run_in_executor(None, self.run_cmd)
# Executed in a different thread; `run_in_executor` submits the
# callable to a thread pool, suspends the awaiting coroutine until
# it's done, and transfers the result/exception back to asyncio.
def run_cmd(self):
time.sleep(3)
return 1
Reading the asyncio documentation, I realize that I don't understand a very basic and fundamental aspect: the difference between awaiting a coroutine directly, and awaiting the same coroutine when it's wrapped inside a task.
In the documentation examples the two calls to the say_after coroutine are running sequentially when awaited without create_task, and concurrently when wrapped in create_task. So I understand that this is basically the difference, and that it is quite an important one.
However what confuses me is that in the example code I read everywhere (for instance showing how to use aiohttp), there are many places where a (user-defined) coroutine is awaited (usually in the middle of some other user-defined coroutine) without being wrapped in a task, and I'm wondering why that is the case. What are the criteria to determine when a coroutine should be wrapped in a task or not?
What are the criteria to determine when a coroutine should be wrapped in a task or not?
You should use a task when you want your coroutine to effectively run in the background. The code you've seen just awaits the coroutines directly because it needs them running in sequence. For example, consider an HTTP client sending a request and waiting for a response:
# these two don't make too much sense in parallel
await session.send_request(req)
resp = await session.read_response()
There are situations when you want operations to run in parallel. In that case asyncio.create_task is the appropriate tool, because it turns over the responsibility to execute the coroutine to the event loop. This allows you to start several coroutines and sit idly while they execute, typically waiting for some or all of them to finish:
dl1 = asyncio.create_task(session.get(url1))
dl2 = asyncio.create_task(session.get(url2))
# run them in parallel and wait for both to finish
resp1 = await dl1
resp2 = await dl2
# or, shorter:
resp1, resp2 = asyncio.gather(session.get(url1), session.get(url2))
As shown above, a task can be awaited as well. Just like awaiting a coroutine, that will block the current coroutine until the coroutine driven by the task has completed. In analogy to threads, awaiting a task is roughly equivalent to join()-ing a thread (except you get back the return value). Another example:
queue = asyncio.Queue()
# read output from process in an infinite loop and
# put it in a queue
async def process_output(cmd, queue, identifier):
proc = await asyncio.create_subprocess_shell(cmd)
while True:
line = await proc.readline()
await queue.put((identifier, line))
# create multiple workers that run in parallel and pour
# data from multiple sources into the same queue
asyncio.create_task(process_output("top -b", queue, "top")
asyncio.create_task(process_output("vmstat 1", queue, "vmstat")
while True:
identifier, output = await queue.get()
if identifier == 'top':
# ...
In summary, if you need the result of a coroutine in order to proceed, you should just await it without creating a task, i.e.:
# this is ok
resp = await session.read_response()
# unnecessary - it has the same effect, but it's
# less efficient
resp = await asyncio.create_task(session.read_reponse())
To continue with the threading analogy, creating a task just to await it immediately is like running t = Thread(target=foo); t.start(); t.join() instead of just foo() - inefficient and redundant.