Deployment

Checklist

composer install --no-dev --optimize-autoloader
php naluz migrate
php naluz route:cache
php naluz view:clear

Setting	Production value
`APP_ENV`	`production`
`APP_DEBUG`	`false` (the default). Never enable it in production
`APP_KEY`, `JWT_SECRET`	strong, secret, set in the environment
HTTPS	enabled. HSTS and `Secure` cookies then switch on automatically
`storage/`	writable by the web server user, not web-accessible
document root	`public/` only

Environment

Set variables in the platform (systemd, Docker, your host’s panel) rather than shipping a .env. Real environment variables take precedence over .env. Never commit .env.

Web server

nginx + PHP-FPM

server {
    listen 443 ssl http2;
    server_name example.com;
    root /var/www/my-app/public;

    location / {
        try_files $uri /index.php$is_args$args;
    }

    location ~ \.php$ {
        include fastcgi_params;
        fastcgi_param SCRIPT_FILENAME $document_root/index.php;
        fastcgi_pass unix:/run/php/php-fpm.sock;
    }

    location ~ /\.(?!well-known) { deny all; }
}

Apache

public/.htaccess is included (needs mod_rewrite). Point DocumentRoot at public/.

Real client IP

Behind a load balancer or CDN, restore the real client address at the web server (nginx real_ip_module, Apache mod_remoteip). The rate limiter and logs use REMOTE_ADDR.

PHP settings

Enable OPcache. In production set opcache.validate_timestamps=0 and reload PHP-FPM on each deploy.
Keep expose_php off and display_errors off.

Caches

Command	When
`php naluz route:cache`	every deploy (controller routes only; closures cannot be cached)
`php naluz view:clear`	every deploy, because compiled templates are trusted in production
`php naluz model-cache:flush`	after out-of-band database changes, if you use model caching

Queue workers

Run workers under a supervisor and restart them on every deploy:

[program:naluz-worker]
command=php /var/www/my-app/naluz queue:work --queue=default --sleep=3 --max-jobs=500 --memory=128
numprocesses=2
autostart=true
autorestart=true
stopwaitsecs=60
user=www-data

A worker finishes its current job on SIGTERM when pcntl is installed. Keep queue.retry_after above your slowest job.

Scheduler

Add one cron entry:

* * * * * cd /var/www/my-app && php naluz schedule:run >> /dev/null 2>&1

Logging in production

Use LOG_CHANNEL=production (daily files plus Slack for critical errors), or stderr with JSON format in containers. See Logging.

Containers

Set LOG_CHANNEL=stderr, pass configuration as environment variables, mount storage/ as a volume (or use shared Redis for cache, sessions and queues) and run php naluz run:server only for development, never in production. Use PHP-FPM with nginx.

Zero-downtime tips

Run migrations that are backward compatible with the previous release.
Restart queue workers after the new code is in place.
Clear the view cache and rebuild the route cache as part of the release step.