Help needed: Using deffered scheduling feature

[MarkoBox]

Hello, Thanks for the library :)

I'm trying to create recuring job which would execute every N days. Looking at the docs I found this:
https://repid.aleksul.space/1.0.0rc3/user_guide/deferred_scheduling/

Example code i had in mind is:

from repid import Job
from repid.router import Router, RouterDefaults
import settings

router = Router(defaults=RouterDefaults(queue=settings.rmq_name))


@router.actor
async def delete_stuff_in_db():
    print("delete_stuff_in_db")


def enqueue_delete_stuff_in_db():
    await Job(
        "delete_stuff_in_db",
        settings.rmq_name,
        timeout=timedelta(seconds=30),
        deferred_by=timedelta(days=7)
    ).enqueue()


if __name__ == "__main__":
    enqueue_delete_stuff_in_db()

Would running this on single container start suffice, or is there something more I'm missing?
In case there are several instances of app running how would this behave. Would i need to make sure only one is running it?

Thanks in advance for help :)

[aleksul]

Hi 👋

or is there something more I'm missing?

Unfortunately, your example won't run as it is. To enqueue a Job, you will need to fix 2 things:

1. Make enqueue_delete_stuff_in_db function async, otherwise you won't be able to do await
2. Open a connection, as described in https://repid.aleksul.space/1.0.0rc3/user_guide/initial_setup/

In case there are several instances of app running how would this behave. Would i need to make sure only one is running it?

Short answer: it depends. In an ideal world, you would set an id to your Job, which will make it unique. This behavior is described here.

However, not all brokers are supporting this functionality (e.g. current Redis implementation does support it, but RabbitMQ - doesn't). You can, of course, create your own synchronization primitive on top of it.