Production Runtime Profile

This profile is designed for backend-only deployment with queue and scheduler separation.

1) Runtime Topology

• nginx: edge HTTP server
• app: PHP-FPM API service
• horizon: Redis queue consumer supervisor
• scheduler: Laravel scheduler loop
• pulse-worker: ingest worker for Pulse stream processing
• mysql: relational database
• redis: cache + queue backend (recommended)

Use docker-compose.production.yml for baseline orchestration.

Production compose only publishes edge ports by default:

• nginx on 8080 by default, configurable via APP_HTTP_PORT
• reverb on 6001 by default, configurable via REVERB_PUBLIC_PORT

mysql and redis stay internal to the Docker network unless you explicitly add host port bindings.

1.5) GHCR Release Image

Stable release tags publish a multi-arch app image to:

• ghcr.io/obsidianlabs-io/obsidian-admin-laravel:<tag>

Supported platforms:

• linux/amd64
• linux/arm64

Tag strategy:

• 1.2.0: immutable full release tag
• 1.2: latest patch in the current minor line
• 1: latest stable minor in the current major line
• latest: stable non-prerelease release only

Use the fully versioned tag for production rollout and rollback control.

Pull example:

docker pull ghcr.io/obsidianlabs-io/obsidian-admin-laravel:v1.2.1

Minimal container boot example:

docker run --rm -p 8080:8000 \
  -e APP_ENV=production \
  -e APP_DEBUG=false \
  -e APP_KEY=base64:Q2qE5A3yM4tQvL3X0yr7M5m4r2m40fX9zCw1Q2m3N4o= \
  -e CACHE_STORE=array \
  -e SESSION_DRIVER=array \
  -e QUEUE_CONNECTION=sync \
  -e AUDIT_QUEUE_CONNECTION=sync \
  -e LOG_CHANNEL=stderr \
  -e LOG_STACK=stderr \
  ghcr.io/obsidianlabs-io/obsidian-admin-laravel:v1.2.1

This minimal run path is only meant to validate image boot and health. For database-backed production usage, prefer docker-compose.production.yml or docker-compose.octane.yml.

Compose consumption example:

services:
  app:
    image: ghcr.io/obsidianlabs-io/obsidian-admin-laravel:1.2.1
    pull_policy: always

Launch with:

docker compose -f docker-compose.production.yml -f docker-compose.image.yml up -d

This lets you keep the repository-provided nginx/mysql/redis/horizon/scheduler topology while consuming a published immutable app image.

Release workflow verification notes:

• the published GHCR app image is a PHP-FPM runtime image, so the release workflow validates cold boot by checking container startup and running php artisan about --only=environment
• route-level HTTP health checks still require an edge/runtime topology such as docker-compose.production.yml or docker-compose.octane.yml

If you want to probe the published image through the real HTTP path, use the published app image with the compose override above and then call:

curl http://127.0.0.1:8080/api/health/live

That route-level probe is already covered in CI by the production Docker smoke job.

2) Start Stack

docker compose -f docker-compose.production.yml up -d --build

Then initialize the app:

docker compose -f docker-compose.production.yml exec app php artisan key:generate
docker compose -f docker-compose.production.yml exec app php artisan migrate --force
docker compose -f docker-compose.production.yml exec app php artisan db:seed --force

3) Health Probes

• Liveness: GET /api/health/live
• Readiness: GET /api/health/ready
• Full diagnostics: GET /api/health

Readiness returns 503 only when critical checks fail.

4) Supervisor Alternative (VM/Bare Metal)

Supervisor templates are included:

• deploy/supervisor/laravel-queue-worker.conf
• deploy/supervisor/laravel-horizon.conf
• deploy/supervisor/laravel-scheduler.conf
• deploy/supervisor/laravel-pulse-worker.conf

Use these when you are not deploying with Docker.