Deploy the complete product

The v0.8.0 candidate includes 16 API services, an AI worker and the React application. Start the media stack and Hermes client separately. Releases defines published versions and image availability; a candidate branch is not a release.

Prerequisites

• Linux, Docker Engine 24+, Compose v2. Start capacity planning around 8 CPU / 16 GB, then load-test your actual workload.
• A TLS domain, strong random secrets and tested database/upload backup and recovery.
• Configure APP_ENV=production, JWT, PostgreSQL, Redis and internal tokens from .env.example. Do not use example passwords.
• Bind service ports to loopback; route public requests through HTTPS to Traefik.
• Configure your own AI, official-account, SMS and telephony providers. Included source code does not activate these integrations.
• Pay currently provides only a mock channel for isolated tests/demos. WeChat Pay and Alipay adapters are not implemented; merchant credentials alone cannot enable real payments. Keep PAY_DEMO_MODE disabled in production.

Build from source

git clone https://github.com/SuperiorChuo/gopherforge.git
cd gopherforge
cp microservices/.env.example microservices/.env
# Edit security and integration settings first.
make compose-up

This creates the shared network, starts the separate infrastructure stack, and uses the compose-fallback profile for the application. The gateway defaults to http://localhost:8000; add TLS for production. Initial credentials are admin / admin123; change the password immediately.

Published images

The release workflow derives its matrix from Compose: 17 unique main-application images (16 APIs and frontend). The worker and migrate job reuse API images. Media and client delivery are separate. Dual-architecture releases started with v0.3.0, but always verify the actual manifests and public commit SHA for each release.

Use this only after v0.8.0 images are published:

cd microservices
export IMAGE_PREFIX=ghcr.io/superiorchuo/gopherforge/go-admin-kit
export IMAGE_TAG=v0.8.0
docker compose --profile compose-fallback pull
docker compose --profile compose-fallback up -d --no-build

Optional components

docker compose --profile compose-fallback --profile monitoring up -d
docker compose --profile compose-fallback --profile tracing up -d
docker compose -p go-admin-kit-infra -f docker-compose.infra.yml --profile storage up -d

The complete deployment guide includes Nginx/HTTPS and independent media/client instructions. Do not expose administrative dashboards publicly.

Upgrade and recover

• Pin the version and full public commit SHA; back up and rehearse on an isolated copy.
• Preserve published SQL paths and bytes. Complete-product upgrades append forward-only migrations; never run production down/reset.
• Application rollback is not schema rollback. Restore the tested, matching database and application versions rather than simply selecting an old image.
• Health checks do not replace API, worker, UI, live-monitoring and required external-provider acceptance.
• Same-host isolation requires independent networks, volumes, container names, ports and credentials. Never reuse production data.

See Upgrades, Observability and Release validation.