Deployment and Operations

The production stack separates frontend, API, worker, PostgreSQL, Redis, object storage, and gateway responsibilities. Long-running generation and rendering belong to the worker.

Operating principles

• Liveness remains healthy when optional downstream systems fail.
• Readiness reports degraded dependencies.
• Every response carries a request identifier.
• Migrations run before a new release serves traffic.
• Database and object-storage backups are both required.
• Provider errors are normalized and credentials are never logged.
• A healthy old container is not proof that current source is deployed.
• Ambiguous paid-provider outcomes enter reconciliation instead of receiving an unsafe refund.

IMAGE PLACEHOLDER - Production topology Replace with Internet/CDN -> TLS proxy -> Caddy -> frontend/API, plus worker, PostgreSQL, Redis, object storage, email, payment, AI, and social providers. Recommended size: 2000 x 1200 px.

After deployment, verify container health, HTTPS, API readiness, migrations, worker connectivity, storage, email, and a safe authenticated journey.