Using Multiple Outputs in a Notebook Process

Below is a polished, professional Knowledge Base article version of your content, structured for clarity, consistency, and production usage.

The Notebook Process in Syntasa enables you to productionize data science workflows by integrating notebooks directly into application dataflows. A key capability of this integration is the ability to define and manage multiple outputs from a single notebook execution.

This feature allows one notebook to generate multiple datasets while also preserving the executed notebook itself for inspection and auditing.

Defining Multiple Outputs

When configuring a Notebook Process node on the workflow canvas, you can specify more than one output destination. These outputs typically fall into the following categories:

  • Notebook Output
    A specialized output node that captures the full executed state of the notebook, including all cells and their results.
  • Dataset Outputs
    Standard Syntasa datasets that are populated by your notebook logic and can be consumed by downstream workflow nodes.

Configuration Steps

To configure multiple outputs for a Notebook Process:

  1. Drag a Notebook Process node onto the workflow canvas
  2. Open the configuration panel and select the source notebook
  3. Navigate to the Output tab
  4. Add one or more datasets
    • Each dataset added appears as a connected output node on the canvas

This configuration defines the managed destinations that your notebook will write to during execution.

Using Output Parameters in Your Code

To ensure your notebook writes data to the correct, managed locations, Syntasa automatically injects output parameters into the notebook at runtime.

Location Parameters

For every output dataset defined in the Notebook Process configuration, a corresponding location parameter is injected using the following convention:

  • Primary Output
    @location
  • Additional Outputs
    @location_<dataset_name>
    (for example: @location_processed_events)

These parameters resolve to cloud storage paths managed by Syntasa.

Example Usage (Python)

# Syntasa injects this value at runtime
output_path = @location_processed_events

# Write results to the managed dataset location
df.write.format("parquet").mode("overwrite").save(output_path)

Using these injected parameters ensures:

  • Output data is written to the correct dataset
  • Downstream workflow nodes can reliably access the results
  • Dataset lifecycle management is handled by the platform

Viewing Outputs After a Job Run

After a job containing a Notebook Process completes, you can inspect all outputs directly within the Syntasa UI.

Notebook Output Node

Each Notebook Process is automatically connected to a Notebook Output node, which represents the executed version of the notebook.

How to View

  • Click the Notebook Output node on the workflow canvas

Available Tabs

  • Output
    Displays a full preview of the executed notebook, including:
    • Code cells
    • Printed output
    • Tables and visualizations generated during execution
  • Details
    Provides execution metadata, including:
    • User who triggered the job
    • Execution completion time
    • Source notebook name
    • Execution environment (Development or Production)

Dataset Previews

For each dataset output produced by the notebook:

  1. Click the corresponding Dataset node connected to the Notebook Process
  2. Open the Preview tab to view a sample of the written data

Syntasa also automatically computes and displays dataset metrics such as:

  • Row count
  • Data size

These metrics are available in the Status or Metrics sections of the job execution details.

Error Handling and Logs

If a notebook execution fails:

  • The Notebook Process node is marked with an ERROR status
  • Detailed execution logs are available through:
    • Job Run History
    • Execution Logs in the Orchestrator module

The logs capture:

  • Standard output (stdout)
  • Standard error (stderr)

These details help diagnose runtime errors, data issues, or logic problems within the notebook.

Summary

Using multiple outputs in a Notebook Process allows a single notebook to:

  • Generate multiple managed datasets
  • Preserve the executed notebook for review
  • Integrate cleanly with downstream workflow components

By leveraging injected location parameters and built-in previews, Syntasa enables scalable, traceable, and production-ready notebook-driven data pipelines.