Browse documentation

Executions

Inspect workflow runs and troubleshoot failures.

Every time a workflow runs, Spark records an execution. The Executions tab in the Workflow section is where you watch progress, confirm success, and trace a failure to the step that caused it.

The executions list

Open Workflow, then the Executions tab. Each row is one run, with columns for Workflow Name, Context, Progress, Status, Created, and Updated. Use Refresh to pull the latest state, or the search box to find a run by workflow name or status.

The Executions list with Filter, Refresh and search, and rows showing Workflow Name, Context, Progress and a Succeeded status.
The Executions list: one row per run, with progress and status.

The Status badge tells you where a run stands:

StatusMeaning
not startedQueued, has not begun
in progressCurrently running
succeededFinished, all steps passed
failedStopped on an error
pausedHeld, for example on a review step
awaiting reviewWaiting on a human decision

A run that shows awaiting review is paused on a Human Review step. See Human-in-the-loop for how to clear it.

Inspect a single run

  1. In the executions list, open the row menu and choose View Details.
  2. The run opens on the canvas alongside the Execution Summary panel.
  3. The Run Details section shows the Run ID, Last Updated, Status, Steps Completed as a completed-over-total count, and Duration.
  4. The Action Details section lists every action. Each shows its Status, a Status Reason, and a Duration.
An execution row's menu with View Details and Copy ID, and a disabled Unblock item.
Open a run's row menu and choose View Details.
The Execution Summary panel with Run Details (Run ID, Last Updated, Status Running, Steps Completed 2/3) and Action Details listing each action.
The Execution Summary: run details and a per-action breakdown, beside the canvas.

To dig into one step, click its node on the canvas to view that action's output and configuration. For an Ask Agent step, the details also show an Agent activity section above the output, with the agent's reasoning and tool steps and, when the model reports it, the token usage for the step.

Read action statuses

Within a run, each action carries its own status: Pending, Running, Succeeded, Skipped, Omitted, Failed, or Error. A red Failed or Error badge marks the step that broke, and its Status Reason usually explains why.

Recover from a failure

For a run whose status is failed, the row menu offers two recovery options:

  • Retry Failed Actions re-runs only the steps that failed, in place, keeping the same execution.
  • Resubmit Workflow creates a new execution and re-runs every action from scratch. Spark asks you to confirm before it does.

Both apply only to a failed run. A run that's paused or awaiting review hasn't failed, so it isn't rerun: you move it forward by resolving its review in the Interventions tab, not with a recovery action. See Human-in-the-loop.

A failed execution's row menu with View Details, Resubmit Workflow, Retry Failed Actions and Copy ID.
A failed run's menu adds Resubmit Workflow and Retry Failed Actions.

Copy ID copies the execution ID, which is useful when raising a support request.

If a scheduled or file-based workflow never appears in Executions, it may not have triggered yet. Check the Trigger States tab: a state of pending means Spark is still waiting on required inputs, while ready, triggered, and expired show how far the trigger got.

Where to go next