Documentation by ClonePartner

Start typing to search.

Using Envoy

Jobs and job templates

View Markdown

Browse Job Templates, create a Job by binding Connectors and arguments, run all or selected tasks, and edit job configuration.

Jobs and job templates

A Job Template is a reusable DAG of Tasks with connector type slots — a blueprint. A Job is an instantiated copy of a template with stored Connectors bound to those slots, ready to execute. Operators work mostly with Jobs; templates are authored by implementers (see Job templates and DAGs).

Screen overview: Job Templates list

Open Templates in the sidebar. The page header reads Job Templates ("Reusable orchestration recipes with connector type slots.") with a refresh button, an Import button, and a New Template button. A search box ("Search templates...") and a Status filter (Active / Archived) sit above the table.

Columns: a selection checkbox, Name (sortable), Description, Status, and Created (sortable). Each row's ⋮ menu offers Export, Archive or Restore, and Delete permanently. Selecting rows switches the toolbar to a bulk-actions bar with Export. Deleting a template asks for confirmation: "This will permanently delete '<name>' and all its jobs. This cannot be undone."

Clicking a row opens the template detail page, which shows the template's status badge, its connector slot chips, Created/Updated dates, and these sections:

  • Connector Slots — a table of the template's typed slots.
  • Task DAG — a table with Task Key, Task, Type, and Dependencies columns describing each node.
  • Dashboard templates — dashboard YAML snapshotted onto each Job created from this template, with an Add dashboard button.
  • Jobs — every Job instantiated from this template, with a New Job button.

Header actions are Edit, Create Job, and a ⋮ menu with Archive/Restore and Delete permanently.

Screen overview: Jobs list

Open Jobs in the sidebar. The page header reads Jobs ("Instantiated job templates with bound connectors.") with a refresh button and a View Templates button. A search box ("Search jobs...") and a Status filter (Active / Archived) sit above the table.

Columns: Name (sortable), Template, Description, Status, and Created (sortable). Each row's ⋮ menu offers Archive or Restore and Delete permanently ("This will permanently delete '<name>' and all its run history."). Clicking a row opens the Job detail page.

To create a job from a template

  1. Open Templates and select the template.
  2. Select Create Job. The Create Job dialog opens ("Configure connectors and arguments to create a runnable job from this template.").
  3. Enter a Name (required; defaults to the template name) and optionally a Description.
  4. Under Connector Mapping, pick a stored Connector for every slot. Each select is filtered to Connectors of the slot's type, and a slot is pre-filled when exactly one matching Connector exists. See Connectors and connected accounts for creating Connectors first.
  5. Under Arguments (shown only when the template declares arguments), fill each field. Required arguments are marked with an asterisk; fields with defaults are pre-filled. Some arguments render searchable pickers (for example integrated-account or MongoDB record pickers) instead of plain text inputs.
  6. Select Create Job. The button stays disabled until the name, every connector slot, and all required arguments are filled. On success you land on the new Job's detail page.

To run a job

  1. Open the Job's detail page.
  2. Select Run All to execute the whole DAG, or check individual rows in the Tasks section and select Run <N> Selected to run a subset. With a selection active, a Skip dependencies checkbox appears — by default, upstream depends_on tasks run first; skipping runs only the checked tasks.
  3. The browser navigates to the new Job Run's page, where you can watch progress. See Job runs.

Tasks in the Utilities section (nodes typed command or validation) each have their own Run button for one-off execution.

Job detail sections

  • Bound Connectors — a Slot / Connector table showing which stored Connector fills each template slot.
  • Configuration — the Job's job_config values, shown in a read-only JSON viewer. Select Edit JSON to edit the values in place and Save (removing keys asks for confirmation). Open dashboard jumps to the first attached Dashboard, and Configure with AI opens the AI assistant with the config as context. Tasks read these values at runtime, and each run snapshots them.
  • Utilities and Tasks — the DAG nodes, split by type. Each row shows the node key, its description, a pencil edit button, and a Run button.
  • Dashboards — Dashboards attached to this Job, with an Add dashboard button. The dialog asks for a Name, Description, and a Start from source: Blank starter, a copy of an existing job dashboard, or a dashboard template.
  • Recent Runs — the Job's run history (#, Status, Started, Duration, Error); rows open the Job Run page.

The header's ⋮ menu offers Archive (or Activate for an archived Job) and Delete permanently.

To customize a task for one job

A Job can override a single DAG node's task YAML without changing the shared Task library. The override is keyed by node, so two nodes using the same base task can diverge independently.

  1. On the Job detail page, select the pencil button on a task row. The editor opens at Edit task: <key>, with the base task name and pinned version in the subtitle.
  2. Edit the YAML and select Save. The node now shows a Customized badge on the Job detail page.
  3. To discard the override, open the editor's ⋮ menu and select Reset to original. The confirmation states: "This removes the per-job override for '<key>'. Future runs will use the pinned task version from the task library."

Run history still records the pinned library task and version, even though the override YAML is what executed.