Jobs

The Jobs page is the operational hub for everything that runs against a model—scans, discovers, rule executions, dataset dumps, loads, pumps, and pipeline runs. It presents a live list of completed activity together with anything waiting in the queue or scheduled for the future.

┌───────────────────────────────────────────────────────────────┐
│ Jobs Page                                                     │
│ ┌─────────────┬────────────┐   ┌────────────────────────────┐ │
│ │ Last Jobs ▣ │ Scheduled │   │ New Job ⊕ • Sort ▾        │ │
│ └─────────────┴────────────┘   └────────────────────────────┘ │
│ ┌───────────────────────────────────────────────────────────┐ │
│ │ Status | Name | Type | Rule | Created By | Info | Actions │ │
│ │   ●    | tap→sink dump | …                               │ │
│ │   ○    | nightly discover | …                            │ │
│ └───────────────────────────────────────────────────────────┘ │
└───────────────────────────────────────────────────────────────┘

UI area	What it shows	How to interact
Tabs (`Last Jobs`, `Scheduled Jobs`)	Two quick filters backed by the same table component. `Last Jobs` shows recent executions. `Scheduled Jobs` pins the table to jobs with status `scheduled` or `queued`.	Switching tabs updates the `tab` query param and refreshes the table with the right filters.
Toolbar (`New Job`, `Sort`)	Primary actions stay on the right. `New Job` opens the New Job modal. `Sort` exposes presets (Activity, Name, Type, User) that map to URL `sortKey/sortOrder`.	Opening the modal requires model-level permissions. Sort selections update the query params so the page is deep-linkable.
Jobs table	Responsive table fed by `Jobs.jobs()` store queries. Displays job metadata, status, and context-aware actions.	Column headers expose filters, row cells carry quick links (job log, rule details, model).

Jobs table columns

Column	Sample value	Details & interactions
`Status`	`running`, `completed`, `failed`, `queued`, `scheduled`	Uses `JobProgress` to render a spinner, success, or failure icon. In the Scheduled tab every row forces status `scheduled`. Filterable with the status pill menu.
`Name`	`tap-to-s3 (2024-05-06 01:00)`	Click to open the job detail page with logs and stream output. The underlying route keeps the current log level filter.
`Model` (project-wide view only)	`Customer Data`	Only rendered when browsing from the project Jobs section. Links back to the owning model via `JobTapColumn`.
`Type`	`dump`, `load`, `pump`, `discover`, `scan`	Filter menu auto-populated from allowed job types (`JobType`). Types control available actions—for example only dump/load/pump jobs can be “Restarted”.
`Rule`	`Anonymize PII`	When present, links to the rule definition (`/rules/:id`). Jobs that generated YAML expose a download action via the `Actions` menu.
`Created By`	`jane.doe`	Renders the user avatar/name responsible for launching or scheduling the job.
`Info`	`Started: 12:21 • Duration: 00:03:18` or `Next: 07/24/2024 22:00`	Shows execution timestamps (`JobWhen`, `JobElapsed`). In the Scheduled tab it switches to “Next run” and “Last run” badges.
`Actions`	`⋮ Cancel • Rerun • Restart • Delete • Download rule as YAML`	Context menu adapts to status and permissions. Cancel applies to `running/scheduled/queued`. Restart appears when the job failed or has entity errors. Delete hides for active jobs.

Quick status reference

Status	What it means	Typical next action
`queued`	Accepted by the worker, waiting to start.	Cancel if the window passed, otherwise leave it.
`running`	Currently streaming entities.	Monitor via job detail; cancel if needed.
`completed`	Finished without errors.	Optionally `Rerun` to repeat.
`failed`	Finished with errors (`hasEntityError` toggles restart availability).	Use `Restart` to retry only failed entities, or `Rerun` for a full execution.
`scheduled`	Has a `nextRunAt` timestamp but has not entered the queue yet.	Edit schedule via `Rerun` (which opens the modal pre-filled) or cancel.

Scheduling and execution modes

Mode	How to configure	Behavior
Run now	Choose Run now in the Schedule panel (default).	Job enters the queue immediately with status `queued → running`.
Schedule	Select Schedule and pick a datetime.	Adds a one-time `scheduledDate`; appears under `Scheduled Jobs` until executed.
Manual execution (pipeline)	Select Manual execution.	Saves the configuration as a pipeline. No job is launched until a user triggers the pipeline run. Requires a pipeline name before saving.
Repeat (pipeline)	Select Repeat, set `Next run on`, and a `Repeat every` cadence.	Creates a pipeline with recurrence metadata. Upcoming runs display in the Scheduled tab with `nextRunAt` filled.

Manual execution and Repeat both convert the modal into a pipeline flow (Save as Pipeline button). Use them for reusable jobs you want to orchestrate from the Pipelines page.

Creating a job

Use the New Job button to open the configuration modal.
The modal is composed of collapsible panels: From, To, Rule, Load options, optional Transform options, and Schedule.
Each panel summarises its state when collapsed so you can review complex setups at a glance.

All modal fields, dependencies, and defaults are documented in detail in the New Job guide.

Where jobs come from

Jobs are launched automatically or manually from several areas:

Discover — comprehensive scans (including quick re-scans).
Rules — anonymisation or transformation runs.
Datasets — dataset dumps, copies, or merges.
Sinks — loading datasets into destinations.
Pipelines — orchestrated sequences that reuse the same job engine.

Whenever one of those actions is scheduled, its future occurrence also appears in the Scheduled Jobs tab so you have a single calendar of upcoming work.