Scheduler
Platform → Scheduler is a read-only operations hub for platform schedule health: KPI tiles, a live calendar, recent tick history, and the schedule event catalog.
What the page shows
- Overview — live KPIs (grace backlog, detention candidates, overdue sync schedules, failed jobs, tick counts, durations), health callouts, and a day/week calendar of expected firings with overdue sync overlays. The overview refreshes automatically about every 45 seconds.
- Recent ticks — filterable table of
platform_scheduler_tick_stats(event, status, time window). - Event catalog — inventory from the platform scheduler catalog (label, cadence, category, scale shape, last tick, runs today).
What you can do
- Watch backlog and tick health without leaving the platform shell
- Filter recent ticks by event, status, and started window
- Browse which schedule events exist and how they scale
What you cannot do here
- Trigger or cancel jobs
- Edit cadences or tenant billing settings
- Mutate tenant data
What to watch for
- Grace backlog due climbing, or oldest age stuck well past ~15 minutes → check queue workers (and
ACCESSORIAL_GRACE_QUEUEif set). - Detention candidate tenants high with no recent detention ticks → check the five-minute detention dispatch / workers.
- Ticks error rising, or missed-tick hints (no tick stats within about two cadences) → open Recent ticks for that event.
- Failed jobs (24h) on default/sweeps → inspect
failed_jobs. - Sync schedules overdue → tenant sync workers /
sync:dispatch-due(separate from grace and detention).
This page never changes billing or detention outcomes — it only shows health.