Skip to main content

DevOps

CI/CD pipelines, infrastructure as code, and containerization standards.

CI/CD Standards​

Every change goes through the same pipeline:

  1. Commit - Push to feature branch
  2. Test - Run unit, integration, and linting checks
  3. Build - Create immutable artifact (Docker image)
  4. Stage - Deploy to staging environment
  5. Manual approval - Team reviews and approves
  6. Deploy - Release to production
  7. Monitor - Track error rates and performance

Pipeline Principles​

  • Fail fast - Run quick checks first (linting, unit tests)
  • Immutable artifacts - Build once, deploy anywhere
  • Environment parity - Staging === Production (as much as possible)
  • Automated rollback - Know how to revert in 5 minutes
  • Audit trail - Every deployment is logged and traceable

Infrastructure as Code​

All infrastructure lives in Git:

├── terraform/
│   ├── main.tf
│   ├── variables.tf
│   ├── outputs.tf
│   └── modules/
├── ansible/
│   ├── playbooks/
│   └── roles/
└── docker-compose.yml

Rules:

  • No manual infrastructure changes
  • All changes via pull request
  • Apply changes via CI/CD pipeline
  • State files are protected and backed up

Docker Standards​

Every service must be containerizable:

Dockerfile Checklist​

  • Multi-stage build (small final image)
  • Non-root user (security)
  • Health checks defined
  • Proper signal handling (SIGTERM)
  • Immutable, pinned base image versions

Example​

FROM node:20-alpine as builder
WORKDIR /app
COPY package*.json ./
RUN npm ci --only=production

FROM node:20-alpine
RUN addgroup -g 1001 -S nodejs && adduser -S nodejs -u 1001
USER nodejs
WORKDIR /app
COPY --from=builder /app/node_modules ./node_modules
COPY . .
HEALTHCHECK --interval=30s --timeout=10s CMD node healthcheck.js
CMD ["node", "index.js"]

Local Development​

Developers should mirror production locally:

docker-compose -f docker-compose-local.yml up

Must include:

  • Application services
  • Database (same version as prod)
  • Redis/cache (if used)
  • Mailhog for email testing
  • Localstack for AWS services

Deployments​

Blue-Green Deployment​

  • Run two identical environments
  • Switch traffic after validation
  • Quick rollback if needed

Canary Deployment​

  • Route small % of traffic to new version
  • Monitor error rates and latency
  • Gradually increase traffic
  • Automatic rollback if threshold exceeded

Never​

  • Rolling restarts without health checks
  • Direct SSH access to production
  • Manual commands in production
  • Database migrations without rollback plan

Documentation​