Quickstart¶

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

Installation¶

pip install nuvom

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.

Dispatch jobs programmatically:

from tasks import add

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

Workers execute jobs in parallel threads:

nuvom runworker

You can configure worker behavior (e.g., count, batch size) via .env. See Configuration for full details.

nuvom inspect job <job_id>

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

To view recent jobs:

nuvom history recent --limit 10

Retry manually from Python:

from nuvom.sdk import retry_job

retry_job("<job_id>")

CLI support for retrying is coming soon.