Technical Documentation
DevOps & CI/CD
Quick search…⌘K
Auto Glass uses a GitOps-based CI/CD pipeline powered by GitHub Actions, Docker, and Kubernetes. Every code change flows through automated linting, testing, security scanning, and a staged canary deployment before reaching 100% production traffic.
CI/CD Pipeline
Loading diagram…
Environments
| Environment | Trigger | URL | Purpose |
|---|---|---|---|
| Local | Developer machine | localhost:3000 | Feature development with hot-reload |
| Preview | Every PR (Vercel) | *.autoglass-preview.vercel.app | Per-PR visual review; read-only staging DB |
| Staging | Merge to main | staging.autoglass.io | Full system integration testing with anonymized prod data |
| Production | Manual approval gate | app.autoglass.io | Live customer traffic; 99.9% SLA |
Docker & Kubernetes Setup
# Service Dockerfile pattern (multi-stage)
FROM node:22-alpine AS builder
WORKDIR /app
COPY package*.json ./
RUN npm ci --only=production
FROM node:22-alpine AS runner
WORKDIR /app
COPY --from=builder /app/node_modules ./node_modules
COPY . .
EXPOSE 3001
USER node
CMD ["node", "dist/server.js"]
# K8s HPA config
apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
name: booking-service-hpa
spec:
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: booking-service
minReplicas: 2
maxReplicas: 10
metrics:
- type: Resource
resource:
name: cpu
target:
type: Utilization
averageUtilization: 60SLIs, SLOs & Alerting
| SLI | SLO Target | Alert Threshold | Escalation |
|---|---|---|---|
| API Availability | ≥ 99.9% per month | < 99.5% in 1h window | PagerDuty P1 |
| API p95 Latency | ≤ 300ms | > 500ms sustained 5min | PagerDuty P2 |
| Error Rate | ≤ 0.1% of requests | > 0.5% in 5min window | PagerDuty P1 |
| Quote Generation Time | ≤ 2 seconds p99 | > 4s p99 | Slack #alerts |
| Dispatch Assignment Time | ≤ 30 seconds | > 60s | Slack #ops |
| DB Replication Lag | ≤ 1 second | > 5s | PagerDuty P2 |
On-Call Rotation: Primary on-call (Monday–Sunday, 24/7) rotates weekly among 4 senior engineers. PagerDuty escalation: 5 min → Primary → 10 min → Secondary → 20 min → Engineering Manager. Runbooks for all P1 scenarios are maintained in Notion under the Incident Response section.