Skip to main content

​
Overview

Each time you run a pipeline, ProDex creates a run record. The record is created at start with empty results and gates, then populated incrementally as the pipeline executes — source data is consumed, transformations execute, gates are confirmed, and result values land. Once execution finishes the record is preserved as part of the pipeline’s history. Runs belong to a pipeline template — the template defines the structure (sources, transformations, outputs, gates), the run is one specific execution against a specific set of inputs.

​
Run States

Each run carries a status that reflects where it is in its lifecycle:
  • pending — queued, not yet started
  • executing — actively running; nodes are being processed
  • completed — every node finished and every gate was confirmed
  • completed_partial — execution finished, but at least one gate was rejected and downstream nodes were left unproduced
A completed_partial run still has useful output. The reachable nodes’ values are saved on the run record — review them and re-run later with corrected inputs or template logic. The unreachable nodes simply have no values for that run.

​
Triggering a Run

Open the pipeline you want to run from the Pipelines workspace, then click Run Pipeline in the top bar. Dexter executes the pipeline and pauses at each gate to ask for confirmation in the chat panel before downstream nodes proceed.

​
What’s in a Run

A run record captures:
  • Inputs consumed — the specific source files or integration extracts that fed the run
  • Results produced — the values that landed at each output node, with their full audit lineage (derivation_explanation, derivation_code, source file references)
  • Gate confirmations — every gate’s evidence and your decision (confirm or reject), along with the confidence level
Click any output node on the canvas (in the run view) to open the derivation drawer for that result — it shows the produced value, the SQL or Python that derived it, the source file it came from, and a plain-English explanation.

​
Reviewing the Runs List

From the runs list you can:
  • See which pipeline each run belongs to and its status
  • Open a run to review its produced results, gate confirmations, and audit lineage
  • Compare result values across runs to see how outputs shifted — whether from updated source data or changes to the pipeline template
  • Delete runs you no longer need