behave-django

behave-django copied to clipboard

Hi, my tests worked fine so far with Channels 1, using the manual integration of behave. Now I'm trying the upgrade to Channels 2, and I don't know how to start the workers anymore. Until now I started some workers as Threads in before_all, but with Channels 2 I only get the following error, when I'm trying to do that:

RuntimeError: There is no current event loop in thread 'Worker'

What can I do, any hints?

@proofit404 helped to deal with the original issue #16 for Channels 1. He, and of course @andrewgodwin, may help us shed light into your problem. -- Artem? Andrew? ping! :bell:

In the meantime, can you please provide specific code samples of your setup to make us understand the specifics of your situation?

This is the relevant part of the environment.py, that was working with Channels 1:

from channels.worker import Worker

WORKER_COUNT = 2


class WorkerThread(threading.Thread):
    def __init__(self, channel_layer):
        super().__init__(name="Worker")
        self.channel_layer = channel_layer

    def run(self):
        worker = Worker(channel_layer=self.channel_layer, signal_handlers=False)
        worker.ready()
        worker.run()


def before_all(context):
    for _ in range(WORKER_COUNT):
        worker = WorkerThread(context.channel_layer)
        worker.daemon = True
        worker.start()

I tried to get it working with Channels 2 - this is the current code, that throws the error:

from channels.worker import Worker
from channels.routing import get_default_application

WORKERS = [
    ["trigger", "ctrl", "device", "script"],
    ["trigger", "ctrl", "device", "script"],
]


class WorkerThread(threading.Thread):
    def __init__(self, channel_layer, channels):
        super().__init__(name="Worker")
        self.channel_layer = channel_layer
        self.channels = channels

    def run(self):
        worker = Worker(get_default_application(), self.channels, self.channel_layer)
        worker.run()


def before_all(context):
    for worker_channels in WORKERS:
        worker = WorkerThread(context.channel_layer, worker_channels)
        worker.daemon = True
        worker.start()

I made some progress, the workers seem to run now with that code:

class WorkerThread(threading.Thread):
    def __init__(self, channel_layer, channels):
        super().__init__(name="Worker")
        self.channel_layer = channel_layer
        self.channels = channels

    def run(self):
        loop = asyncio.new_event_loop()
        asyncio.set_event_loop(loop)
        logger.info(loop)
        worker = Worker(get_default_application(), self.channels, self.channel_layer)
        loop.run_until_complete(worker.run())
        loop.close()

I don't know much about that asyncio stuff, so maybe it's not a good solution. I also now have the problem, that some channel messages get lost and aren't handled, and I get occassional errors like "Two event loops are trying to receive() on one channel layer at once!".