Skip to content

Quickstart

This guide walks you through installing Nuvom, defining your first task, scheduling it, and running workers all in under 5 minutes.

Installation

pip install nuvom

1. Define a Task

Tasks are regular Python functions decorated with @task:

# tasks.py
from nuvom.task import task

@task(retries=2, retry_delay_secs=5, timeout_secs=3, store_result=True)
def add(x, y):
    return x + y

The decorator enables retry logic, timeouts, and lets you dispatch with .delay() or .map().

Nuvom uses static AST-based discovery to find task definitions without executing your code.

Run once:

nuvom discover tasks

This generates .nuvom/manifest.json to speed up worker startup and avoid runtime imports.

3. Submit a Job Immediately

Dispatch jobs programmatically:

from tasks import add

job = add.delay(5, 7)
print(job.id)

4. Schedule a Task

Tasks expose .schedule() for deferred or recurring execution. Scheduling works seamlessly across workers and backends.

from datetime import timedelta, datetime, timezone
from tasks import add

# Run at a specific time (2038 iykyk)
add.schedule(5, 7, at=datetime(2038, 1, 19, 3, 14, 7, tzinfo=timezone.utc)) 

# Run once after 30 seconds
add.schedule(5, 7, in_=timedelta(seconds=30))

# Run every 5 minutes (interval scheduling)
add.schedule(2, 3, interval=300)

# Cron-style: every day at midnight UTC
add.schedule(1, 2, cron="0 0 * * *")

Start the scheduler service:

nuvom runscheduler

Workers will automatically execute due jobs.

5. Run a Worker

Workers execute jobs in parallel threads:

nuvom runworker

You can configure worker behavior (count, batch size, scheduler backend, etc.) via .env. See Configuration for details.

6. Inspect Job Status

nuvom inspect job <job_id>

This shows result, error, traceback, retries remaining, and timestamps.

To view recent jobs:

nuvom history recent --limit 10

7. Retry Failed Jobs

Retry manually from Python:

from nuvom.sdk import retry_job

retry_job("<job_id>")

CLI support for retrying is available in the next release.