I'm still testing this out locally but I figured I'd go ahead and throw up a PR.

This would add support for cron syntax for scheduling tasks at certain times versus just on certain intervals.

For example, I might want something to run specifically "at 3AM" versus "every 24 hours". This preserves the default intervals for timedelta objects in config.

Hello, first i'd like to thank you for this module. I would like to have several periodic tasks. I've thought about grouping them in a specific channel? so the would be all together. But as i see in the code, it's about a single task per channel name. In order to support different message types in one channel, there should be 2 cycles here https://github.com/rajasimon/beatserver/blob/58feddc3dd2b20ba12a99f51c9a70d6cc7c49b5b/beatserver/server.py#L30 Outer one should cycle through channel name/config(key-value), and inner one should cycle through message types. Or there would be one cycle, but config shoud be a list instead of dictionary(maybe a list of tuples?), so that there could be several items with the same channel name and maybe the same message type.

After installed with command pip3 install -U beatserver, OS can not recognize command 'beatserver' on OSX. BTW, beatserver is a websocket server which can continuely send message to clients browser frontend if I didn't make any misunderstanding?

These changes give beatserver Channels 2 support without dropping Channels 1 support. Admittedly the modifications are a bit hacky due to my inexperience and the difference between Channels 1 and Channels 2 APIs, so maybe it would make more sense to fork the project into a Channels-1 version and a Channels-2 version to keep the code clean...

In addition to the changes in these files, the beatserver.py file requires additional lines to work correctly in Channels 2 (no changes needed if the user wants to keep working with Channels 1) which should be documented in the README file. This is an example beatserver.py file that works with Channels 2:

import os from datetime import timedelta from channels.layers import get_channel_layer

os.environ.setdefault("DJANGO_SETTINGS_MODULE", "PROJECT_NAME.settings")

BEAT_SCHEDULE = { 'send-every-5-seconds': { 'channel_name': 'CHANNEL_NAME', 'schedule': timedelta(seconds=5), 'message': {'type': 'type_of_message', 'message': 'message_text'} }, }

channel_layers = get_channel_layer()

and the server needs to be initialized with the following command:

beatserver PROJECT_NAME.beatconfig:application

Putting all this information in the beatconfig.py file is necessary to get the channel_layer. This is due to the fact that in Channels 2 if you try to import the PROJECT_NAME.asgi file as was done for Channels 1 you get a conflict between a the reactor imported in Server.py and an asyncioreactor used by Daphne. Daphne notices the conflict and proceeds to uninstall automatically the synchronous beatserver reactor which breaks beatserver, so I had to move the necessary information to get the channel_layers to the beatconfig.py file.... I know that this quite inelegant and you'll probably want to find a better solution before merging my changes, but I thought you'd prefer to at least have some "first working solution" you can improve upon :-).

hey, I need to run 2 different task with beatserver, I use django2.x and channels2.x, in consumers.py I write a class BeatServer as:

class BeatServer(SyncConsumer):
    def task_one(self, message):
        print("task 1 #############")
        data = {"task_one": "ok"}
        room_group_name_one = "room_one"
        async_to_sync(self.channel_layer.group_send)(room_group_name_one, {
            "type": "chat.message",
            "message": data
        })

    def task_two(self, message):
        print("task 2 #############")
        data = {"task_two": "ok"}
        room_group_name_two = "room_two"
        async_to_sync(self.channel_layer.group_send)(room_group_name_two, {
            "type": "chat.message",
            "message": data
        })

and beatconfig.py as below

from datetime import timedelta

BEAT_SCHEDULE = {
    'task-one': {
        'type': 'task_one',
        'message': {'message': 'send successed'},
        'schedule': timedelta(seconds=4)
    },
    'task-two': {
        'type': 'task_two',
        'message': {'message': 'send successed'},
        'schedule': timedelta(seconds=4)
    },
}

then run beatserver as python3 manage.py beatserver I can see task 1 #############, and things were going well on client but task two seens not work, i cannot see any tag, and clients can not receive any messages. and there are no errors raise.

can it be done if more than one tasks with beatserver?

Hey, thanks for the project! I cloned the repo and tried to run the example, but there are no static files, namely, there is no websocketbridge.js file

Currently the mentioned task in the beatconfig execute after the schedule time and repeat but that's okay but make it option to execute the task immediate. I think the immediate execution of the task should be in default mode and make it options to the current behaviour.

Hello, As i understand, spinning up the beatserver is done separately from spinning up my django server. So how would i go about using this in production, if my stack consists of Gunicorn + Daphne (channels and db already configured with redis channel layer )

So for simplicity, lets say that my current deployment script runs this command sudo systemctl restart gunicorn.service && sudo systemctl restart nginx && sudo systemctl restart daphne.service

and in my gunicorn config my ExecStart is gunicorn my_app.wsgi --bind 0.0.0.0:8000 --log-level error --log-file=- --workers 5 --preload

daphne ExecStart ExecStart=/home/ubuntu/env/bin/daphne --bind 0.0.0.0 --port <my_port> --verbosity 2 my_app.asgi:application

how can i also spin up beatserver to play nice with the others ?

This converts the existing code snippets into fenced code blocks and adds language identifiers in order to render them with syntax highlighting.

I've also used bold formatting on the file names above each code block to make them stand out more.

One of the cool thing about ASGI is we can hook with asyncio loop easily and that allows us to implement background task not just to channels but any ASGI framework. I'm going to invest some time around this year to change the code to support for all the asgi applications.