Each step decides what happens when it fails, under If this step fails in its settings. Setting it to try again lets a run get past a service that is down for a moment or busy, without you having to run it again by hand.

Who can do this

Workspace Admins and Editors, in the designer. Viewers see the setting but cannot change it.

Steps

  1. Select Workflows in the bar at the top of the page, and open the workflow.
  2. Select the step on the canvas. Its settings open on the right, on the Settings tab.
  3. Scroll to If this step fails.
  4. Choose one of three:
    • Stop the run — Nothing after this step happens. The run shows as failed. This is the default.
    • Try again up to 3 times, 30 s apart — Then stop, unless it works.
    • Carry on and pass the error along — Later steps get the error in their input and can decide.
  5. If you chose to try again, set Times (2 to 5) and Wait (5, 10, 15, 30, 45 or 60 s). The choice's wording changes to match, such as Try again up to 5 times, 10 s apart.
  6. Publish the workflow so live runs use the new setting. Test runs use it straight away.

What happens next

  • While a step waits to try again, the run's page shows it as Retry 2 of 3, and its error reads Attempt 1 of 3 failed: … Trying again in 30 s.
  • If a try works, the run carries on as if nothing had happened, and the step shows as succeeded.
  • If every try fails, the step fails and the run stops, as with Stop the run.
  • With Carry on and pass the error along, the step still shows as failed, its error ends Carried on, as this step is set to., and the next steps receive each item with the error on it — so a condition can check it, or a message can say what broke.

Good to know

  • A trigger has no such setting — only the steps after it do.
  • Only the step tries again, never the whole run from the top: what already worked is not repeated.
  • Trying again is for things that fail for a moment — a service that is briefly down, a rate limit. A wrong field fails every time, as fast.
  • The wait between tries counts towards the time limits: 15 minutes for a stretch of a live run, 5 minutes for a whole test run.
  • To run a whole run again after you have fixed something, use Run again on the run's page. See Run a workflow again from a past run.

If something goes wrong

What you see Why What to do
Every try fails with the same error The cause is in the settings or the data, not the service. Read the error and the step's Input, and fix the cause. See Fix a failed run.
Stopped after 15 minutes. A run that needs longer should wait between steps. Many tries with long waits used up the run's 15 minutes. Use fewer tries or shorter waits.
The setting cannot be changed You are a Viewer in this workspace. Ask an Admin or Editor.