vercel-workers 0.0.9

Python SDK for Vercel Workers

vercel-workers

Installation

pip install vercel-workers

Django 6.0 django.tasks integration (Vercel Queues)

This repo ships an optional django.tasks backend that enqueues tasks into Vercel Queues and executes them via a queue callback route (similar in spirit to the Celery integration in this repo).

Install

pip install "vercel-workers[django]"

Configure Django

In settings.py:

TASKS = {
    "default": {
        "BACKEND": "vercel.workers.django.backend.VercelQueuesBackend",
        # Optional: restrict which Django queue_name values are allowed.
        "QUEUES": ["default"],
        "OPTIONS": {
            # TaskResult storage (uses Django cache; for production prefer Redis).
            "cache_alias": "default",
            "result_ttl_seconds": 86400,

            # Callback execution settings (visibility/lease).
            "visibility_timeout_seconds": 30,
            "visibility_refresh_interval_seconds": 10.0,

            # Basic retry policy for task exceptions.
            "max_attempts": 3,
            "retry_backoff_base_seconds": 5,
            "retry_backoff_factor": 2.0,

            # Optional publish-time overrides passed through to vercel.workers.client.send()
            # (you can also set env vars like VERCEL_QUEUE_TOKEN / VERCEL_QUEUE_BASE_URL).
            # "token": "...",
            # "base_url": "...",
            # "base_path": "...",
            # "retention_seconds": 86400,
            # "deployment_id": "...",
            # "timeout": 10.0,
        },
    }
}

Define and enqueue tasks

from django.tasks import task


@task(queue_name="default")
def send_email(to: str) -> None:
    ...


result = send_email.enqueue(to="[e-mail]")

Expose the Vercel Queue callback route

You must expose a WSGI/ASGI endpoint that Vercel Queues can POST CloudEvents to:

• WSGI: vercel.workers.django.get_wsgi_app(backend_alias="default")
• ASGI: vercel.workers.django.get_asgi_app(backend_alias="default")

How you mount that depends on your deployment setup (Vercel Python entrypoint / framework).

Notes / limitations

• Queue naming: Django Task.queue_name maps directly to the Vercel Queue name you publish to.
• Priority: not supported (Django Tasks will validate priority is default).
• run_after: supported by delaying the queue message (visibility timeout) when it is delivered early.
• Results: stored in Django cache. With LocMemCache, get_result() is process-local; for real cross-request results use Redis (or another shared cache backend).