Skip to content

Managing schedules

There is one schedule store, and it surfaces everywhere a schedule matters.

Once a flow is saved, the Schedule toggle in the canvas toolbar opens the visual schedule dialog. It includes:

  • An enable switch plus hourly / daily / weekly / monthly / yearly presets.
  • Advanced cadences: arbitrary cron, every-N-minutes, and one-shot runs.
  • Timezone selection (IANA), end conditions (until-date, max runs), and the catch-up policy.
  • A preview of the next firings for the current configuration.

The flow’s Graph properties dialog shows a summary of the active schedule.

A central view lists every schedule along with its flow, cadence, next run, run count, and last status. Each row has its own enable and disable, edit, and delete controls. This is the fleet console. When a dozen flows carry schedules, this is where you see them all.

Template rows show schedule state inline. Each row shows whether a schedule exists, how it is set up, its run count, its last status, and its next run. This means the Load modal also works as a quick schedule audit.

  • The History view filters to scheduled runs (trigger = scheduled). This keeps unattended activity separate from interactive work.
  • Each completed scheduled run raises a notification through the header bell. A failure that fired at 6 a.m. is waiting for you at 9.
  • The schedule’s own run count and last status update with every firing.

Two scheduler controls live in Settings:

  • Pause all is a global switch that stops the scheduler loop without touching individual schedules.
  • Poll interval sets how often the background task checks for due schedules.

Scheduled runs are unattended, so each one passes a per-run override that disables the interactive destructive-action confirmation gate. There is no one present to confirm. The persisted setting still governs interactive runs. Practical guidance:

  • Review a flow’s destructive surface before you schedule it. Pre-apply verification and the canvas Checks section flag delete and force operations at authoring time.
  • Prefer flows whose failure path is a diagnostic branch, such as downloading logs or sending a notification, rather than a corrective destructive step.

The application core exposes the full management surface. This covers list, get, upsert, delete, preview, enable and disable, plus firing a due schedule. It is reachable from the desktop commands, the server’s invoke bridge, and the CLI, so external tooling can manage schedules the same way the UI does.