Skip to main content
When a workflow doesn’t do what you expect, the run tells you why. Find the step that went wrong, fix the cause, then retry the run.

Find the problem

1

Open the run

Open the workflow’s Runs tab. Filter by Status to find Failed or Completed with errors runs, or by Run ID if someone shared one.
2

Select the step that went wrong

Select the first step marked Failed or Skipped on the path that didn’t finish.
3

Read the failure, inputs, and log

The Failure message says what broke. Inputs show the values the step actually used. Log shows what happened while it ran, including why it was skipped.
Learn more about run and step statuses

Common issues

  • Check the switch in the editor reads On. Draft and paused workflows don’t run.
  • Compare the trigger’s filters with the ticket. Runs whose filters didn’t match are hidden from the run list, so a missing run usually means a filter excluded the ticket.
  • For triggers that depend on an integration, such as Slack reactions, check that the integration is connected.
  • For a Webhook trigger, send requests to the workflow’s current webhook URL. Only published workflows accept them.
A step is skipped when a True, False, or approval-outcome connection into it didn’t pass, or its Optional Filters didn’t match. Everything after it on that path is skipped too.
Learn more about skipped steps
  • Read the Failure message on the step.
  • Check Inputs for empty or unexpected values. A dynamic value that resolves to nothing is a common cause.
  • Check that the integration the step uses is connected and has the permissions the action needs. In the builder, actions show Setup Required when their integration isn’t connected.
  • Check whether the external service is having an outage.
  • Compare the step’s Inputs with what you expected.
  • Open the dynamic value and check it points at the right step and field.
  • Check the trigger’s Outputs to confirm the field exists for this kind of ticket. Some values exist only on some tickets, such as email details on tickets created from email.
  • Compare each step’s runtime to find the slow one. Calls to external services are the usual cause.
  • An HTTP Request waits up to its Timeout, 10 seconds by default and 45 seconds at most.
  • Automatic retries add a few seconds between attempts.
  • Check wait steps. A run stays Running while it waits.
Runs in progress are stopped when someone selects Stop & Publish, deactivates the workflow, or deletes it. Cancelled runs can’t be retried. Trigger the workflow again instead.

Automatic retries

Ravenna retries a step automatically when it fails for a temporary reason, such as a timeout or a server error. A step gets up to 3 attempts in total, a few seconds apart, and each attempt can run for up to 5 minutes. Errors that won’t go away on their own, such as a missing required value or a record that doesn’t exist, aren’t retried.The step’s details show Attempts when it ran more than once. A step that fails on its last attempt is marked Failed, and you can retry the run.

Retry a run

Retry a run to recover from a temporary failure, to pick up a fix you published, or to run the same trigger data again. Retry appears in the row menu of runs that are Failed, Completed with errors, or Completed. Pending, running, and cancelled runs can’t be retried.Every retry starts with the original run’s trigger data, so it acts on the same ticket. It creates a new run, linked to the original.
Retry from failure and Retry from beginning use the version the original run used, so they don’t include changes you’ve published since. To pick up a fix, choose Retry with latest version.
Retrying repeats each rerun step’s side effects. A step that sent a message sends it again.
1

Open the run history

Open the workflow and select the Runs tab.
2

Open the row menu

Find the run and select Retry from its row menu.
3

Choose a retry policy

Select the option that matches your goal, then select Retry.
4

Follow the new run

Select View run in the confirmation, or find the new run at the top of the Runs tab.

Retry through the API

Retry a run from the Ravenna API with POST /workflows/runs/{runId}/retry:
The response contains the new run’s runId.
If you combine retryFromFailure with a newer versionId, steps that completed in the original run aren’t rerun, even if you changed them. To run a changed step again, retry from the beginning on the new version.

Skip wait steps

A workspace admin can end a Wait or Wait Until step early. Open the run, select the step while it is Waiting, and select Skip Wait. The workflow continues immediately, and this can’t be undone.Use it when the reason for the wait has already been handled, or to keep testing without waiting for the timer. Other wait steps, such as Wait for Approval and Wait for Message, continue when someone responds on the ticket.
Last modified on November 18, 2026