View and manage pipelines¶
After adding a pipeline, click the Pipelines tile. Each listing shows the pipeline name, any description, and Draft or Locked status. Select a pipeline to inspect its graph, configuration, and run history.
The page opens to Workflow view for the selected pipeline—one of three available tabs:
| Tab | Description |
|---|---|
| Workflow view | An interactive directed acyclic graph (DAG) for inspecting task-level telemetry and output. |
| Details | Pipeline identity, tasks, current input, and the bound image. |
| Runs | A chronological log of previous runs. |
The Images tab lists container images. For more information, see Manage pipeline images.
Draft and Locked pipelines¶
A pipeline has two states.
- Draft. Logic, inputs, and image can be modified; use this state while iterating.
- Locked. Logic and environment are immutable; recurring schedules can target only Locked pipelines.
For the concept definitions, see Draft and locked pipelines.
To lock a pipeline and attach a schedule, see Lock and schedule.
Workflow view¶
Workflow view renders the task-call graph as a DAG. Each node is a @dr.task function. Independent branches run in parallel; downstream tasks wait for their inputs. For the selected run, the view shows:
- The run timestamp, ID, and status (for example, Run completed).
- Each task node with its status and run time.
- A task list on the right.
Use the zoom controls to inspect the graph.
Task list¶
From the task list on the right, click and expand a task. The expanded task has three tabs for additional information.
The Details tab shows when and how this task ran, the resource bundle it used, and the function source.
| Field | Description |
|---|---|
| Triggered by | How the run started (for example, On-demand). |
| Started | When the task started. |
| Run time | How long the task took. |
| Resource bundle | Compute size for the task (for example, S, 1 CPU / 512MB RAM). |
| Code | The @dr.task function source for this node. |
Return value is what that @dr.task function returned for the selected run—the object that DataRobot passes to downstream tasks as arguments. Nothing is shared through files or memory except this value. The tab provides a JSON preview of the result (in the example, prepare_data shows [[140, 1], [130, 1], …]). Click Download raw value for the full stored object (including objects that do not preview cleanly, such as a fitted scikit-learn model).
The Logs tab shows console output from the task for the selected run. Click View logs on a task in the list to open this tab. The tab provides:
- stdout (standard output): Text the task writes with
print(). - stderr (standard error): Warnings, exceptions, and Python tracebacks. When a task fails, the traceback is usually here.
Refresh, copy, or download from the icons on the pane.
Details¶
The Details tab summarizes pipeline configuration: identity and timestamps, the task list, the current input payload, and the bound image.
The following table describes the fields on Details.
| Field | Description |
|---|---|
| Pipeline ID | Unique identifier for the pipeline. Click the copy icon to copy the ID. |
| Latest version | Locked version of the pipeline. None yet (draft) appears when the pipeline has not been locked. |
| Python version | Python runtime for the pipeline (for example, 3.12). |
| Created | When the pipeline was created. |
| Modified | When the pipeline was last changed. |
| Tasks | Each @dr.task in the DAG and the resource bundle assigned to that task (for example, S, 1 CPU / 512MB RAM). |
| Current input | YAML payload used by the most recent run. Each run can override this payload. |
| Current execution image | Image bound to the pipeline: Name, Version, Build status, image Python version (or Platform default), and the packages list. |
Next steps¶
After reviewing a pipeline, run it, configure a schedule, or return to the list to add another pipeline.
- Run and schedule pipelines: Run the pipeline, lock it, and review run history and logs.
- Add a pipeline: Create another pipeline from Python source, inputs, and an image.
- Manage pipeline images: Create or rebuild the container image used at run time.