deploy-guidelisted

# Step 6: Deploy (ปล่อยของ) **Habit**: H1 — Be Proactive | **Anti-pattern**: Deploying directly to production without staging ## Process 1. **Check existing deploy infrastructure**: Read deploy scripts, CI/CD configs, docker-compose files, Makefiles — understand what's already automated. 2. **Classify the deploy type before planning rollout**: - `image build` — artifact/image changes; deploy the new artifact through staging first. - `mounted config/template` — runtime reads a file mounted from host/volume; plan config render, sync, service reload/restart, and verification. - `Swarm config` — `docker config`/secret object changes; plan config object replacement plus service update. - `full stack` — compose/stack topology changes; plan broader blast radius and rollback. - `source-of-truth drift` — repo/template says one thing but live runtime reads another; reconcile before deploy. - `force-update only` — no artifact/config change, but scheduler needs a controlled restart; still needs health/rollback checks. - `production canary / capacity change` — cloud/provider-managed infrastructure change where the operator can pick a canary but the provider may mutate a different resource; plan reconciliation gates. - `no deploy` — truly non-runtime work. 3. **Staging first** (ALWAYS when deploy type has runtime impact): ``` Step 1: Build artifacts (images, bundles) Step 2: Deploy to staging Step 3: QA on staging (health check, feature verify, smo