Runbook
What to do at 3am. Written for someone tired and unfamiliar.
- Format.md
- Length76 lines
- Includesnone
The source
76 lines of Markdown, with nothing to install. Copy it, or open the template inside Gnomon and render it as it is.
# Runbook: <Service / Feature>
- **Owner team:** <team>
- **On-call rota:** <link to PagerDuty / Opsgenie schedule>
- **Last updated:** YYYY-MM-DD
## Purpose
What this service does, in one paragraph. Why it matters if it's down.
## Architecture Refresher
Quick diagram + 3–4 bullets so a fresh on-call can orient in under a minute.
Link to the full HLD for details.
- Entry points: …
- Critical dependencies: …
- Datastores: …
## Dashboards
- **Service health:** <link>
- **Latency / errors:** <link>
- **Infra:** <link>
## Alerts
| Alert | Severity | Meaning | First response |
|--------------------------------|----------|--------------------------------------|--------------------------------------------------|
| `HighErrorRate` | P1 | 5xx rate > 5% for 5m | Check recent deploys; consider rollback |
| `HighLatency` | P2 | p95 > 1s for 10m | Check downstream; check DB connections |
| `QueueDepthGrowing` | P2 | Backlog growing for 15m | Scale workers; check for poison messages |
| `HealthCheckFailing` | P1 | Liveness failing on > 1 instance | Check logs; restart pod; escalate if persistent |
## Common Incidents
### Symptom: 5xx spike after deploy
1. Check the most recent deploy in <CI link>.
2. If correlated, roll back: `<command>`.
3. Post in #<channel>; open incident.
### Symptom: queue backlog growing
1. Check worker count and CPU: <dashboard link>.
2. If healthy, scale: `<command>`.
3. If unhealthy, check DLQ for poison messages.
### Symptom: latency rising, errors flat
1. Check downstream service status pages.
2. Check DB slow query log.
3. Check connection pool saturation.
## Useful Commands
```bash
# Tail logs
…
# Scale workers
…
# Drain a node
…
```
## Escalation
| Level | Who | When |
|-------|-------------------------|-----------------------------------------------|
| L1 | Primary on-call | All P1/P2 alerts |
| L2 | Secondary on-call | No ack within 10 min, or L1 requests help |
| L3 | Team lead / EM | P1 not mitigated within 30 min |
| L4 | Director / incident cmd | P1 user-visible > 1h, or data loss suspected |
## Post-Incident
- File a [post-mortem](./post-mortem.md) within 5 business days for any P1.Render this offline
This template ships in Gnomon and renders on your machine, with no account and nothing sent to a server. The browser editor is free and needs no install.
Others in Architecture document templates
- ADR — Architecture Decision RecordContext, decision, consequences. The alternatives section is the valuable one.
- RFC — Request for CommentsFor proposing a change and inviting disagreement before building.
- HLD — High-Level DesignHigh-level design: the shape, the constraints, the risks.
- Tech Spec / LLDLow-level design. What an implementer needs that the HLD does not say.
- NFR CatalogueThe requirements that get discovered late and expensively.
- Discovery / Spike ReportFindings from a time-boxed investigation, including the dead ends.
- Post-Mortem / Incident ReviewBlameless incident review. Timeline first, conclusions second.
- Solution One-PagerA solution summary for people who will not read the HLD.