Runs and troubleshooting

Every execution is recorded. Runs lists them all; opening one shows each step in order with its input, its output, how long it took, and any error.

Run states

• Queued — accepted, waiting for a worker.
• Running — in progress.
• Waiting — parked at a delay, an approval, or a subflow. It isn't stuck and isn't using a concurrency slot.
• Succeeded / Failed — finished.
• Canceled — you stopped it.

When something fails

Open the run and find the first failed step. The error is translated into something actionable where we can — a 401 means the connection needs reconnecting, a 422 usually means a required field is empty or the wrong shape. The step's Input shows exactly what was sent, which is normally where the problem is.

Fix the workflow, then use Replay to run it again with the original trigger data. Replays are linked to the run they came from, so you can see the whole history. Where recovery is available you can also resume from the failed step instead of repeating the successful ones — which matters when the earlier steps had real side effects.

Stopping a run

Cancel run stops a queued, running or waiting run. A queued or waiting run stops immediately; a running one finishes its current step and then stops, so a half-finished API call is never left in limbo.

Being told about failures

Under Notifications you can have YakPipes email you when a run fails, with a quiet window so one broken integration doesn't produce a hundred emails. A daily or weekly digest summarises activity instead. Individual workflows can also POST failures to a URL of your choosing.

Running over history

A workflow normally only sees things that happen after you turn it on. To process what is already there — old orders, existing contacts, last year's tickets — use Run over history (linked from Workflows). Pick the workflow, the app operation that lists the records, and a ceiling.

Each record starts one ordinary run, so a backfill counts against your run quota like anything else. When the quota window is spent the backfill pauses itself and resumes later rather than failing, so starting a large import on a small plan is fine — it just takes longer. Records already processed are remembered, so stopping and resuming never double-processes anything, and you can stop a backfill at any time.

Retention

Run history is kept for 90 days by default; you can change that under Settings → Data lifecycle, where you can also export everything your account holds.