Deployment

Build and run

cargo build --release -p example-app
./target/release/example-app

The compose stack builds the app from the workspace Dockerfile and exposes a /health liveness probe on port 8000.

docker-compose up -d
docker-compose down

Infrastructure services

Production configuration

Two rules to internalize before shipping:

1. Environment variables always win over TOML files. The config loader merges config/*.toml first, then applies the process environment last — so secrets belong in the environment, not in files.
2. Destructive migrations are gated. migrate:fresh is refused in production unless --force is passed explicitly.

APP_ENV=production APP_KEY=... DATABASE_URL=... cargo run --release

Run migrations as a release step, then bring the app up:

cargo artisan migrate --force

Background workers

Queue workers and the scheduler are separate processes in production:

cargo artisan queue:work --queue=default
cargo artisan schedule:run

For recurring dispatch, drive schedule:run from an external timer or a long-running supervisor, and use the pause/resume commands during deploys to avoid double dispatch:

cargo artisan schedule:pause
# deploy
cargo artisan schedule:resume

Release checklist

• cargo xtask ci passes (fmt, clippy, deps, lines, DAG cycles)
• cargo test --workspace passes
• cargo deny check and cargo audit are clean
• APP_KEY and all secrets come from the environment
• Migrations run with --force and are reversible
• Health probe on /health matches the orchestrator configuration