sunnydata-infrastructure

繁體中文說明：此技能整合 docker-patterns 與 deployment-patterns，涵蓋從本機容器化到正式環境部署的完整工作流程。

Infrastructure

Overview

Single skill for "how code runs" — from local Docker to production deployment. Merged from docker-patterns (container dev workflow) and deployment-patterns (CI/CD, release strategy).

Activate when:

• Containerizing an application (Dockerfile, .dockerignore)
• Setting up Docker Compose for local development
• Designing multi-container or multi-network architectures
• Establishing a CI/CD pipeline
• Planning a deployment strategy (Rolling, Blue-Green, Canary)
• Implementing health checks or Kubernetes probes
• Preparing a production release checklist

Core Principles

• One canonical multi-stage Dockerfile per language. The dev stage is used by Compose; the production stage ships to production — minimal, non-root, exact pinned base versions, HEALTHCHECK built in, secrets never baked into image layers.
• 12-Factor configuration. All config via environment variables, never hardcoded; validate at startup and fail fast if config is wrong.
• One health endpoint per service. Simple /health liveness plus a detailed readiness check that probes real dependencies (database, cache, external APIs).
• Every deploy has a tested rollback path. Backward-compatible migrations, tagged previous artifacts, feature flags that disable without a deploy.
• Build once, promote the artifact. The same image digest moves dev → staging → production through evidence gates; only environment variables and secrets differ per environment. Never rebuild for production.

Workflow

1. Containerize. Read references/dockerfile-patterns.md when writing or reviewing a Dockerfile — it holds the canonical Node.js/Go/Python multi-stage patterns, the shared .dockerignore, and container security hardening (non-root, capability drops, secret management).
2. Local development. Read references/compose-local-dev.md when setting up or debugging Docker Compose — standard web app stack (app/db/redis/mailpit), override files, service discovery and network isolation, volume strategies, and the debugging command reference.
3. Pipeline and release. Read references/cicd-deployment.md when building CI/CD or planning a release — GitHub Actions pipeline, deployment strategy details, health endpoints and Kubernetes probes, environment validation, environment promotion gates (dev → staging → production, migration ordering), rollback procedures, and the production readiness checklist.

Deployment Strategy Decision

Scenario	Strategy
Standard backward-compatible change	Rolling
Critical service, instant rollback required	Blue-Green
High-risk change, need real-traffic validation	Canary
Database schema change (additive only)	Rolling with migration guard

One-line summaries (full traffic diagrams and trade-offs in references/cicd-deployment.md):

• Rolling (default): replace instances one at a time; zero downtime, but two versions run simultaneously — changes must be backward-compatible.
• Blue-Green: two identical environments, atomic traffic switch; instant rollback at the cost of 2x infrastructure during deployment.
• Canary: route a small traffic percentage to the new version first; catches issues with real traffic but requires traffic-splitting and monitoring.

Pipeline Stages

PR opened:
  lint → typecheck → unit tests → integration tests → preview deploy

Merged to main:
  lint → typecheck → unit tests → integration tests → build image → deploy staging → smoke tests → deploy production

Production Gate

Before any production release, run the production readiness checklist in references/cicd-deployment.md (application, infrastructure, monitoring, operations). For the security portion, load sunnydata-security for the technology-specific checklist (secrets, CORS, rate limiting, auth, security headers, CVE scanning).