Run lifecycle
Runs move through the following statuses:| Status | Description |
|---|---|
| Queued | Run is waiting to start |
| Running | Pipeline steps are actively executing |
| Completed | All steps finished successfully |
| Failed | A step encountered an unrecoverable error |
| Review | Run is paused, waiting for human review |
| Cancelled | Run was manually cancelled by a user |
| Awaiting Children | Run is paused while child runs (from document chunking/splitting) complete. See document chunking |
Step executions
Each node in the pipeline produces a step execution within the run. A step execution records:- The node type and configuration used
- Start and end timestamps
- Status
- Input data received from the previous step
- Output data produced for the next step
- Processing time
- Error details (if failed)
Step execution statuses
| Status | Description |
|---|---|
| Pending | Step is waiting to execute |
| Running | Step is actively processing |
| Completed | Step finished successfully |
| Failed | Step encountered an error |
| Skipped | Step was skipped (e.g., conditional logic or pipeline configuration) |
| Retrying | Step failed and is being retried automatically |
| Review | Step is paused for human review |
Document chunking
When a document is split into chunks (e.g., a multi-page PDF split into individual pages), DocPipe creates child runs for each chunk. The parent run enters the Awaiting Children status until all child runs complete. Each child run processes its chunk independently through the pipeline. The parent run tracks:- Child runs — individual processing runs for each chunk
- Chunk index — the position of each chunk in the original document
- Chunk label — a descriptive label for the chunk