Jobs

A job runs a notebook, a script or a command to completion, once or on a schedule. Each run is a pod of its own, with your home folder attached, and leaves its log and outputs in your home folder.

Job templates

Template	What it runs
Python jobs	Jupyter notebooks (.ipynb, with papermill), marimo notebooks (.py), Quarto documents (.qmd, .ipynb), or a command
R jobs	Jupyter notebooks with the R kernel, or a command
Julia jobs	Jupyter notebooks with the Julia kernel, or a command
Custom command	A script you type, run with bash -lc

A plain script runs as a command, for example python report.py, Rscript report.R or julia report.jl. A command runs without a shell; use Custom command for pipes and other shell features. A .py file picked as a file is treated as a marimo notebook.

Admins also see Backups, the template of the backup and restore jobs.

Create a job

1. Open Jobs and press New job, or try one of the sample jobs.
2. Pick a template, then under What runs choose File (and a kernel for notebooks) or Command.
3. Add parameters if the job takes any (text, whole number, number or yes/no). Commands get them as environment variables JAVO_PARAM_<NAME>.
4. Add environment variables and secrets, an environment if your admin set up Nebi, and a size.
5. Under When should it run? choose Run once (it starts as soon as you create the job) or On a schedule.

Schedules

The schedule builder offers every minute, every few minutes, hourly, daily, weekly (pick the days), monthly, or a custom cron expression. Pick a time zone: it defaults to your browser’s, and the builder previews the next run times and warns about daylight-saving changes (or lets you use UTC).

Hidden behind the options:

• When a run is still going: skip the new run, run both, or replace the old run.
• Retries after a failure.
• Time limit, such as 1h.

On the job page, Pause schedule and Resume schedule stop and restart scheduled runs; Run now starts one at any time.

Runs, logs and outputs

The job page lists each run with when it started, its outcome, how long it took and its output. A run’s log is followed live while it runs, then read from its folder. A running run can be cancelled. After a failure, Re-run with same inputs starts it again, and Open executed notebook opens the notebook as it ran, with its outputs.

Each run writes to ~/jobs/<job>/<run>/ in your home folder:

• run.log: the run’s log.
• output.ipynb and output.html for notebooks, output.html for marimo.
• result.json: the run’s result.

Size, environment variables and secrets

A job’s size and environment variables can be changed on the job page after it is created. Secrets reach the run by reference, never in its spec: a scheduled run gets the values current at its start, so a changed secret applies from the next run; a run already going keeps what it started with.

Failure notices

When your admin has set up notifications, a failed run sends a notice to the recipients the admin configured and to the job owner’s verified email. The job page shows what the job notifies on.