Running Pipelines
Learn how to execute pipelines and understand the execution lifecycle.
Once you have built a pipeline on the canvas, you can run it to process data through your nodes. Salt executes your pipeline in the cloud and streams results back to you in real time.
How to Run a Pipeline
Save your pipeline
Make sure your Pipeline is saved. Salt auto-saves as you work, and you can save manually with Cmd+S / Ctrl+S.
Run the Pipeline
Select Run in the Pipeline header. This sends your Pipeline to the execution engine.
Watch it run
Each node highlights as it begins processing and updates when it finishes. You can see the overall progress in real time.
View results
When execution completes, click on any node to inspect its output in the Properties panel. You can also press R to open the Run History panel, which shows a log of past executions.
Execution Lifecycle
Every Pipeline run follows this lifecycle:
- Queued -- Your Pipeline is submitted and waiting to start.
- Running -- Nodes execute in dependency order. Each node reports its own status as it runs.
- Completed or Failed -- The run finishes successfully or stops with an error.
If a node fails during execution, the error is reported on that specific node. Other independent branches of your pipeline may still complete successfully.
Execution Order
Salt determines the execution order automatically based on your edges:
- Nodes with no incoming edges (or only widget values) run first.
- A node runs only after all of its upstream dependencies have completed.
- Independent branches of the pipeline run in parallel when possible.
You don't need to specify execution order manually -- it is derived entirely from the edge graph.
Viewing Past Executions
Use Run History to inspect past executions and their outputs. This is separate from Pipeline history, which contains saved editor snapshots you can preview or restore. To recover an earlier Pipeline state, see Pipeline version history.
Press R or select the History control in the right-side header group to review your previous manual and agent-triggered runs. Its tooltip reads Open runs. Each entry can show:
- When the execution started
- Whether it succeeded or failed
- How long it took
- Whether an agent triggered it
- How many outputs it created
Open a run to review its outputs, add a personal note or favorite, inspect errors, and use the actions available for that run. See Run History for filtering, bulk actions, output downloads, and deletion rules.