Job
A Renku job is a non-interactive run that executes a command in the background using a project's Environment and compute resources.
A job is submitted from a Job launcher (see Launcher) and runs without a browser-based interface. Inside the job, all of the project's Code repositorys are cloned and Data connectors are mounted, as long as you provided any required access credentials when submitting.
Each job submission is identified by a submission ID — a short name you choose (or accept from the auto-generated default) that must be unique among your jobs on that launcher. You use the submission ID to tell jobs apart when several are running at once.
A job has access to compute resources (CPU, GPU, RAM, and storage) determined by the resource class you select at submit time. For more information, see Resource Pools & Classes.
The jobs you submit are always private to you. Jobs are not shared between users.
Job lifecycle
A job moves through states such as Starting, Running, Completed, and Errored. While a job is starting or running, you can view its logs or cancel it. When a job has completed or errored, you can view its logs and then Dismiss it to remove it from your project and dashboard.
Dismissing a job permanently removes it from the RenkuLab interface. You will not be able to reopen its logs after dismissing it.
Completed and errored jobs are typically removed automatically after about 24 hours if you have not dismissed them. This retention period may change in the future.
When to use a Job
Use a Job when you want to run a script or batch process in the background — for example, model training, data processing, or any long-running command that does not require an interactive notebook or editor.
If you need an interactive environment for exploration or development, use a Session instead. See How to handle long training runs if you are deciding between a Job and workarounds inside an interactive session.