Download the PHP package provydon/laravel-scale without Composer

Laravel Scale

Scale your Laravel app with one install: it comes with Laravel Octane (FrankenPHP), a production-ready Docker setup, and a stateless web + worker layout that easily runs your Laravel app on Render, Laravel Cloud, Fly.io, Railway, AWS ECS & EKS, Google GKE & Cloud Run, and other container platforms.

Why You Need Laravel Scale

Most Laravel apps run on a single server with fixed PHP-FPM workers—often just one or two processes serving all requests. That works until traffic spikes (launch, campaign, viral moment). Then requests queue, response times blow up, users hit timeouts, and revenue is lost. You can scale up the server, but you still have the same one or two processes and there's always a CPU and memory ceiling.

Cloud platforms scale differently: they replicate your app across many instances and distribute traffic automatically (horizontal autoscaling). Laravel isn't ready for that out of the box. You need containerization, web + worker separation, stateless sessions, external cache/queues/storage, and a runtime like Octane. Getting that right takes time.

Laravel Scale gives you that autoscaling setup with just one install—no Kubernetes knowledge required. Your app will run in Docker, deploy to platforms like Render, AWS, GCP, Railway, and autoscale with demand instead of collapsing under spikes.

Move from Old Fixed server capacity → New Cloud-native autoscaling. If you're building something that could spike, go viral, or serve millions and want to stay in Laravel. This helps your Laravel app be ready when it happens.

Getting started

In your project folder on your local machine, download the package and run scale:install—that's it.

Next, commit and push the files and folders the command added to your project—they're needed for autoscaling:

• docker/
• .dockerignore
• app/Providers/ForceHttpsServiceProvider.php
• app/Http/Middleware/ForceHttpsMiddleware.php
• bootstrap/providers.php
• bootstrap/app.php
• config/octane.php

Deploying on a Cloud Platform (eg Render.com)

1. Create a Web Service (Docker)

1. In Render Dashboard, click New + → Web Service.
2. Connect your repo (GitHub/GitLab).
3. Configure:
   • Name: e.g. myapp-web
   • Region: choose one
   • Environment: Docker
   • Dockerfile Path: docker/Dockerfile (required—set in Advanced if not visible).
   • Instance Type: pick size (e.g. Starter or higher).
   • Leave the rest default or set if you need to.
4. Environment variables (Add Environment Variable):
   • APP_KEY (generate with php artisan key:generate --show locally).
   • APP_ENV=production, APP_DEBUG=false. If you leave APP_ENV=local or leave it unset, the app may generate http:// asset URLs and the page can appear blank (Mixed Content blocked by the browser).
   • APP_URL set to your production URL with https:// (e.g. https://myapp.onrender.com or your custom domain). Use your own domain; platform defaults like *.onrender.com can cause session issues.
   • APP_NAME: if the name contains spaces, wrap the value in an extra single quotation so it is parsed correctly, e.g. APP_NAME='"Digitalize with AI"'.
   • DEPLOYMENT_TYPE=web.
   • Database: DB_CONNECTION, DB_HOST, DB_PORT, DB_DATABASE, DB_USERNAME, DB_PASSWORD. Use any database (PostgreSQL, MySQL, SQLite, etc.); on Render you can add a PostgreSQL instance and copy its env vars.
   • Stateless: For session, set only these three and remove any other SESSION_* from .env.example and your platform env so Laravel defaults apply (avoids session issues when deployed): SESSION_DOMAIN=.example.com (replace with your root domain, e.g. .yourdomain.com), SESSION_DRIVER=database, SESSION_LIFETIME=120. Set LOG_STACK=single,stderr so logs go to both file and stderr (your platform can capture stderr). Also set CACHE_STORE=database, QUEUE_CONNECTION=database, and if using S3: FILESYSTEM_DISK=s3, AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_DEFAULT_REGION, AWS_BUCKET.
   • Optional—broadcasting: BROADCAST_CONNECTION=pusher (or reverb/ably) and driver-specific env vars (e.g. PUSHER_APP_ID, PUSHER_APP_KEY, etc.) on both Web and Worker.
5. Port: set Port to 8000 (Octane listens on 8000).
6. Save and deploy. Render will build the Docker image and start the web service.

Set APP_URL correctly. Use either the URL your platform gives you (e.g. https://myapp.onrender.com) or your custom domain (e.g. https://app.yourdomain.com). If APP_URL is wrong, the app can load over HTTPS but request assets over HTTP. The browser will block those requests (Mixed Content), and the page will appear blank—e.g. "The page at 'https://yoursite.onrender.com/login' was loaded over HTTPS, but requested an insecure resource 'http://yoursite.onrender.com/build/assets/...'. This request has been blocked." Fix it by setting APP_URL to the exact public URL (including https://) in the Web Service environment variables, then redeploy.

If npm run build fails (e.g. Vite/Wayfinder errors): add a Docker build argument SKIP_FRONTEND = 1 in the service's Environment so the image builds without frontend assets. See docker/README.md for details.

2. Create a Worker Service (optional)

1. New + → Background Worker.
2. Same repo; Environment: Docker.
3. Dockerfile Path: docker/Dockerfile (same as Web).

4. Build Command (optional, for slimmer image):

5. Start Command: leave empty (entrypoint uses DEPLOYMENT_TYPE to start queue + scheduler).
6. Environment variables: same as web (DB, Redis if used, APP_KEY, etc.) plus:
   • DEPLOYMENT_TYPE=worker
7. No port needed. Deploy.

3. Render-specific env vars

Render injects some vars automatically (e.g. RENDER_EXTERNAL_URL, RENDER_INSTANCE_ID). The Docker entrypoint skips copying RENDER_* from host env into .env so they don't overwrite; your app can still read them from getenv() or $_ENV if needed.

4. Database

Use any database Laravel supports: PostgreSQL, MySQL, or SQLite. The Docker image includes all three drivers by default—no Dockerfile edits needed. Set DB_CONNECTION, DB_HOST, DB_PORT, DB_DATABASE, DB_USERNAME, and DB_PASSWORD in your Web and Worker services. On Render, create a PostgreSQL or MySQL instance (or use SQLite for a single instance) and add its connection vars.

Important: make your database reachable and grant access

Your app runs inside containers. For migrations and normal requests to work, two things must be true.

1. The database must be reachable from your containers.
If the DB runs on a separate host (managed service or your own server), it has to accept connections from the internet or your platform's private network. The right port must be open: 3306 for MySQL, 5432 for PostgreSQL. On Render, attaching a managed PostgreSQL or MySQL instance does this for you. For a self-hosted or external DB, open that port in the firewall and allow your platform's outbound IPs (or put the DB and app on a private network).

2. The database user must have access to the database.
The user in DB_USERNAME must be allowed to connect to the host and must have privileges on the database named in DB_DATABASE (e.g. SELECT, INSERT, UPDATE, DELETE, and for migrations CREATE, ALTER, etc.). On managed services this is usually set when you create the DB and user; on your own server, grant the user access to the database and ensure it can connect from the app's network.

If the container fails to start or you see "No open ports": Check the deploy logs. The entrypoint runs migrations and other steps before starting the web server; if migration fails or another step fails, the container exits and no port is opened. The entrypoint prints a clear error (e.g. "ERROR: Migration failed…") and a short hint on common causes (database not reachable, DB user has no access, missing env vars). Fix the cause (open port 3306/5432, grant the user access to the database, or set APPKEY and DB*), then redeploy.

5. Redis or key-value cache (optional)

If you use Redis for cache, session, or queue, add the PHP Redis extension to the Dockerfile first: in the install-php-extensions block, add redis \ on its own line before the other extensions. Then deploy a Redis or key-value cache service (e.g. Render's Redis add-on, Upstash, or similar) and point the web service at it via env configs. Set CACHE_STORE=redis, SESSION_DRIVER=redis, and/or QUEUE_CONNECTION=redis, then add the connection vars (REDIS_HOST, REDIS_PASSWORD, REDIS_PORT, etc.) to your Web and Worker services. No code changes—Laravel reads these from the environment.

6. Custom domain and HTTPS

Use your own domain or subdomain for the app instead of the platform's default (e.g. *.onrender.com). Platform default domains can cause session handling issues and sessions may invalidate unexpectedly. Add your domain in the Web Service → Settings → Custom Domains; Render provides TLS. Set APP_URL to the exact URL your app's DNS points to (e.g. https://app.yourdomain.com). When the app is behind a load balancer, APP_URL must match the public URL or links, redirects, and assets can break.

scale:install adds *`trustProxies(at: '')** and **statefulApi()** tobootstrap/app.phpso that behind a reverse proxy (e.g. Render) the app uses the forwarded protocol and generateshttps://` asset URLs. Without this, the page can load over HTTPS but request CSS/JS over HTTP (mixed content), and the frontend may appear blank.

Contents

• After install
• What it does
• Typical dev journey
• Local development
• Contributing (install from local path)
• Support

After install

1. Octane in production
   scale:install adds or moves laravel/octane into require in your composer.json so the Docker image (which runs composer install --no-dev) gets Octane. The package is in require-dev so production’s composer install --no-dev won’t install it; the published files run in production.

2. Database
   The Docker image includes MySQL, PostgreSQL, and SQLite drivers by default (pdo_mysql, pdo_pgsql, pdo_sqlite). Set DB_CONNECTION and your DB_* env vars (e.g. on Render) and it works. See the “Database” section under Deploying on a Cloud Platform above for essential requirements: your database must be reachable from your containers (correct ports open) and the DB user must have access to the database.

3. Stateless setup
   In Render (and .env.example), set:

   • Session: Remove all existing SESSION_* variables, then set only SESSION_DOMAIN=.example.com (replace with your root domain, e.g. .yourdomain.com), SESSION_DRIVER=database, and SESSION_LIFETIME=120. Leave other session settings as Laravel defaults so sessions work properly in this stateless environment (see docker/README.md for details).
   • Logging: Set LOG_STACK=single,stderr so the platform can capture logs from stderr.
   • Cache: CACHE_STORE=database (or redis).
   • Files: FILESYSTEM_DISK=s3 and AWS_* (or other external disk).
   • Queue: QUEUE_CONNECTION=database (or redis) for the worker service.

4. Docker

   • Web: build with DEPLOYMENT_TYPE=web, expose port 8000.
   • Worker: same image with DEPLOYMENT_TYPE=worker, or build with --build-arg DEPLOYMENT_TYPE=worker for a smaller image.
5. Render
   • Web: Docker service, build from repo, start command = container default (entrypoint), port 8000.
   • Worker: second Docker service, same image, DEPLOYMENT_TYPE=worker, no port.

Why separate web and worker-scheduler services?

You need both a web service and a worker-scheduler service (Background Worker on Render):

• Scheduler: Running the scheduler (schedule:work) on a single dedicated worker avoids race conditions. If every web container ran the scheduler, multiple instances could trigger the same task at once (e.g. duplicate emails or cleanup jobs).
• Queue and HTTP: Running queue:work and the scheduler on the worker keeps background work off the web processes. That way web containers stay focused on handling requests instead of being slowed or blocked by queued jobs and cron.

See docker/README.md (published into your app) for the full stateless checklist, build commands, PHP version (how the image picks PHP 8.2–8.5 and how to pin a version), database (the image includes MySQL, PostgreSQL, and SQLite drivers by default—no Dockerfile edits needed), and backend-only apps (how to remove the Node/frontend stage if your app is API-only).

What it does

scale:install publishes into your app:

• docker/ — Dockerfile (PHP + Node frontend stage; MySQL, PostgreSQL, SQLite drivers; PostCSS/Tailwind support), entrypoint, supervisord-web.conf (Octane), supervisord-worker.conf (queue + scheduler), php.ini
• .dockerignore — keeps build context small
• app/Providers/ForceHttpsServiceProvider.php — forces https:// for URLs in non-local environments (so production works behind a reverse proxy without Mixed Content)
• app/Http/Middleware/ForceHttpsMiddleware.php — runs at the start of the web stack so asset/Vite URLs use https:// (avoids blank or unstyled pages from Mixed Content)
• docker/README.md — stateless checklist (session/cache in DB or Redis, files on S3), PHP version, database (all three drivers by default), backend-only variant

Requirements: PHP ^8.2 (Laravel 13 needs PHP ^8.3 in your app), Laravel ^11.0|^12.0|^13.0, laravel/octane ^2.17 (FrankenPHP). Run composer update in your app to pull compatible versions.

Typical dev journey

1. Local setup – Create your Laravel app, develop as usual (Blade, Inertia, API, etc.).
2. Install once – When ready to deploy: composer require provydon/laravel-scale --dev --with-all-dependencies and php artisan scale:install.
3. Commit – Commit docker/, .dockerignore, app/Providers/ForceHttpsServiceProvider.php, app/Http/Middleware/ForceHttpsMiddleware.php, bootstrap/providers.php, bootstrap/app.php, config/octane.php, and the .gitignore changes. Push to GitHub/GitLab.
4. Stateless config – In .env.example (and your platform’s env), remove all other SESSION_* and set only SESSION_DOMAIN=.example.com (use your root domain), SESSION_DRIVER=database, SESSION_LIFETIME=120; set LOG_STACK=single,stderr; set CACHE_STORE=database, QUEUE_CONNECTION=database. Use S3 for uploads. Add Redis extension to Dockerfile if using Redis. See docker/README.md for the full session/deployment checklist.
5. Platform – Create a Web Service (Docker) and a Background Worker on Render (or similar). Point both at your repo. Set Dockerfile Path to docker/Dockerfile for each. Add a database (e.g. PostgreSQL on Render) and set env vars, deploy.
6. Iterate – Push code; platform rebuilds from the repo. No scale:install in CI—everything is already in the repo.

Local testing (optional): Run docker build -f docker/Dockerfile --build-arg DEPLOYMENT_TYPE=web -t app:latest . and docker run -p 8000:8000 -e APP_KEY=base64:xxx -e DB_*="..." app:latest to test the image locally.

Local development

Use Laravel Octane as usual, e.g.:

Or your existing composer dev / npm run dev setup; the package only adds Docker and publishables for deployment.

Contributing (install from local path)

If you’re developing this package or want to try it from a local clone, in your Laravel app’s composer.json add:

Then run composer update provydon/laravel-scale --with-all-dependencies and php artisan scale:install.

When you upgrade the package (e.g. composer update provydon/laravel-scale --with-all-dependencies), run php artisan scale:install again to publish the latest Docker files and fixes, then commit any changed files.

This will:

1. Publish docker/ (Dockerfile, Dockerfile.backend, entrypoint, supervisor configs, php.ini). Then ask whether your app has a frontend (Vite, Blade with assets, etc.); if no (or you pass --no-frontend), the main docker/Dockerfile is set to the backend-only version (no Node stage). Both variants are in docker/; you can switch later by overwriting Dockerfile with Dockerfile.backend or re-running the install.
2. Publish .dockerignore.
3. Run octane:install --server=frankenphp (unless you pass --no-octane).
4. Update .gitignore so docker/ and config/octane.php are committed (unless you pass --no-gitignore). Update .env.example so only the three stateless session vars are set (unless you pass --no-env-example).
5. If you use Laravel Wayfinder: remove the Wayfinder plugin from vite.config.* (so npm run build in Docker doesn't run PHP) and add !resources/js/routes/, !resources/js/actions/, and !resources/js/wayfinder/ to .gitignore so generated files are committed. Run php artisan wayfinder:generate locally and commit the output before deploying.

Choosing full vs backend-only Dockerfile

The install publishes two Dockerfiles: docker/Dockerfile (full: Node + PHP, builds Vite/assets) and docker/Dockerfile.backend (PHP only, no Node). It then sets which one is the main docker/Dockerfile:

• Interactive: You are asked "Does this app have a frontend in the same repo (React, Vue, Svelte, Vite, etc.)?" — Yes (default) keeps the full Dockerfile; No overwrites docker/Dockerfile with the backend-only version. Press Enter for the default (full stack).
• --no-frontend: Skips the question and uses the backend-only Dockerfile (for API-only apps).
• Non-interactive (e.g. CI): No question is asked; the full Dockerfile is kept.

You can switch later by overwriting docker/Dockerfile with docker/Dockerfile.backend, or re-run scale:install and answer the question to restore the full one.

Options:

• --no-octane – Only publish files; don't run Octane install.
• --no-dockerignore – Don't overwrite .dockerignore.
• --no-gitignore – Don't update .gitignore (ensures docker/ and config/octane.php are committed).
• --no-env-example – Don't update .env.example (keeps existing session vars; install normally strips other SESSION_* and sets only SESSION_DOMAIN, SESSION_DRIVER, SESSION_LIFETIME).
• --no-wayfinder – Skip Wayfinder Vite and .gitignore adjustments.
• --no-frontend – Use the backend-only Dockerfile (no Node/Vite); for API-only apps. Without this, scale:install asks whether your app has a frontend and sets docker/Dockerfile accordingly (full or backend-only).

Support

Enjoying Laravel Scale? Buy me a coffee — it helps keep this project maintained.

License

MIT.