Skip to main content

Documentation Index

Fetch the complete documentation index at: https://septemberai.mintlify.app/llms.txt

Use this file to discover all available pages before exploring further.

This page is the source of truth for the Engine’s on-call rotation — who’s on call, what they do, and how the schedule works. Like every other operational doc, it’s live: edit it as we evolve.

​
Cadence

Currently a weekly rotation, primary + secondary. Each engineer takes one week per N (where N is the number of engineers in the rotation). A primary handles pages first; a secondary is the backup. The handoff happens at Monday 10:00 IST. The off-going on-call runs the handoff checklist; the in-coming on-call acknowledges before the off-going leaves.

​
Tooling

The pager is currently TBD (PagerDuty / OpsGenie / similar). On-call gets paged for any Sev1 or Sev2 alert; Sev3 tickets are created in the issue tracker for next-business-day handling. The rotation schedule lives in the same tool. Engineers are responsible for arranging swaps when they need them; the schedule is the source of truth.

​
What on-call is for

The on-call engineer is responsible for:
  • Responding to pages within 15 minutes. Acknowledge, don’t necessarily resolve. Acknowledging means “I’m aware and starting to work the issue.”
  • Triaging and mitigating. Identify the alert, follow the team’s agreed playbook for that alert class, and surface gaps after the incident closes.
  • Escalating when needed. See the severity definitions.
  • Communicating. Update the incident channel on progress.
  • Filing a postmortem after Sev1/Sev2. See Postmortems.
The on-call engineer is not responsible for:
  • Routine engineering work. On-call hours are for oncall.
  • Code reviews, planning meetings, or anything else that requires focus. If a fire is burning, drop everything else.
  • Resolving the root cause during the page. Mitigate first; root-cause in a postmortem.

​
Coverage

Currently business hours coverage only (10:00 – 19:00 IST, weekdays). Outside business hours, customer-impacting issues queue into the incident channel and are addressed at the start of the next business day. This is a deliberate choice for the current scale. As we onboard production customers with stricter SLAs, we’ll expand to 24/7 coverage, which requires both more engineers and a follow-the-sun rotation.

​
Compensation

On-call hours during business hours are paid at the engineer’s normal rate. There is no on-call premium today. When we expand to 24/7, on-call outside business hours will carry an on-call premium plus straight-time for actual incident hours.

​
Hand-offs

See the handoff checklist. The 5-minute version:
  1. Off-going lists open incidents and their state.
  2. Off-going lists alert noise from the past week.
  3. Off-going lists changes deploying this week.
  4. In-coming acknowledges in the rotation channel.

​
Swaps and exceptions

Engineers can swap weeks freely. Find a willing trade partner, post the swap to the rotation channel, update the schedule. No approval needed. Planned absences (vacations, conferences) should be flagged at least two weeks in advance. The rotation will be adjusted before the schedule is published. Sick days: if you’re sick during your week, message the rotation channel and the secondary takes over. If neither of you can cover, the last engineer who was on-call before you takes over.

​
Onboarding to the rotation

A new engineer joining the rotation:
  1. Reads the latest five postmortems.
  2. Shadows an existing on-call for one week (no responsibility, just watch).
  3. Goes on rotation as secondary for two weeks.
  4. Becomes primary.
Total onboarding time: ~4 weeks.

​
Off-rotation engineers

Engineers not in the rotation can still be paged for high-severity issues that touch their area of code. The on-call engineer makes the escalation call — see severity. This is best-effort, not a contractual obligation. Off-rotation engineers are not on-call.

​
See also