Skip to main content

​
Runbook: Schedule with no members

Symptom: The on-call query for a schedule returns an empty result. Escalation policies that reference the schedule have nobody to page. The UI shows “Nobody on call” for the schedule.

​
Diagnose

1. Check who is currently on call:
An empty on_call array means the schedule has no coverage at this time. 2. Check the schedule configuration: Go to On-Call → Schedules → [schedule name]:
  • Are there any rotation layers? An empty schedule has no layers.
  • Do the layers have participants? A layer with no users produces no on-call.
  • Is the rotation start date in the future? A layer that hasn’t started yet produces no coverage.
  • Is there a gap in the rotation (e.g., the schedule only runs Mon–Fri and it’s Saturday)?
3. Check for overrides: An override that removes the on-call person without a replacement creates a gap. Go to Schedule → Overrides and check the current time window. 4. Check user accounts: If a participant’s Regen account was deleted after being added to the schedule, the slot is empty. The schedule calendar will show the deleted user’s name but they can no longer be paged.

​
Mitigate

Add a temporary override:
  1. Go to On-Call → Schedules → [schedule name] → Overrides
  2. Create an override for the current gap, assigning an available engineer
  3. This takes effect immediately — the next escalation check will find the override
Add a user to the rotation layer:
  1. Go to the schedule → edit the rotation layer
  2. Add the missing participant
  3. The rotation recalculates immediately for future shifts
Add a fallback tier to the escalation policy: If the schedule will have gaps regularly (e.g., weekends only covered by a secondary team):
  1. Go to On-Call → Escalation Policies → [policy name]
  2. Add a final tier with a named individual (not a schedule) as a backstop
  3. This tier fires if all earlier tiers produce no on-call result

​
Resolve

  1. Verify GET /api/v1/schedules/<id>/oncall returns a non-empty result
  2. Create a test incident and confirm the escalation policy pages correctly
  3. Review the schedule for the next 30 days to identify any other coverage gaps

​
Prevention

  • Every schedule should have at least two participants in each rotation layer — a one-person rotation creates a single point of failure
  • Every escalation policy should have a final tier with a named individual (not just a schedule) as a fallback
  • Review schedule coverage monthly — look especially at holiday periods and timezone gaps
  • Set up a /api/v1/schedules/<id>/oncall monitor that alerts if the result is empty for more than 5 minutes