Skip to content
FlowArray help

Nothing happened — no run, no notification

Most silences are a guard doing its job: a schedule armed for its next slot, a payload that matched nothing, a mute you forgot about.

In this article (5)

A schedule did not run

  • Just saved it? A schedule is always armed for its next slot — saving never fires immediately. The dialog shows Next run; that is the truth. Same after changing the time or un-pausing.
  • A slot passed while FlowArray was down? Missed slots are skipped, never caught up — deliberately, so an outage cannot produce a burst of paid runs.
  • Was the previous run still going? Overlapping slots are skipped by default.
  • Is it paused? A paused schedule spends nothing and fires nothing.
  • Check the flow still has a runnable version and that the spend limit is not reached.

A trigger call did not produce a run

Look at Recent deliveries on the trigger first — it records every call and what it did.

  • {"ignored": true} means the payload mapping produced zero rows. That is a free no-op, on purpose, so chatty webhooks cost nothing. Fix the mapping if you expected rows.
  • 404 means the trigger id or secret is wrong — including after a rotate (the old secret dies instantly) or a pause. It also stops working if the person who created it left the workspace.
  • Same run id returned twice means your Idempotency-Key matched an earlier call. That is the duplicate protection working; change the key for a genuinely new event.
  • 429 means the trigger's hourly cap (10 runs by default) or per-call item cap (20) was hit.
  • Check the secret is in the header, not the URL.

"finished": false came back

Only if you used ?wait=. FlowArray holds the response up to 25 seconds; if the run is not done it returns the run id and finished: false. The run is still going — do not treat it as an error and do not retry. Poll the run instead.

A notification did not arrive

In order:

  1. Is the channel on? Settings → Preferences for email and webhook.
  2. Is notify on set to only failures for a run that succeeded?
  3. Is a digest configured? A digest replaces per-run pings with one summary later.
  4. Is the flow muted — by you, or workspace-wide by an admin (its card would show Alerts off)?
  5. Notifications only ever cover your own runs.

A run is stuck "running"

Long is not stuck: video genuinely takes minutes per clip, and a Repeat loop multiplies a step's duration by its attempts. The run monitor shows what is in flight and whether a key is cooling down.

If a worker restarts mid-generation, FlowArray re-polls the paid job it already holds rather than submitting a new one, so recovery does not double-bill. A job it can no longer reach is parked as needs review instead of being silently retried.