Skip to content

Latest commit

 

History

History
580 lines (411 loc) · 22.9 KB

octane.md

File metadata and controls

580 lines (411 loc) · 22.9 KB

Laravel Octane

Introduction

Laravel Octane supercharges your application's performance by serving your application using high-powered application servers, including Open Swoole, Swoole, and RoadRunner. Octane boots your application once, keeps it in memory, and then feeds it requests at supersonic speeds.

Installation

Octane may be installed via the Composer package manager:

composer require laravel/octane

After installing Octane, you may execute the octane:install Artisan command, which will install Octane's configuration file into your application:

php artisan octane:install

Server Prerequisites

Warning Laravel Octane requires PHP 8.1+.

RoadRunner

RoadRunner is powered by the RoadRunner binary, which is built using Go. The first time you start a RoadRunner based Octane server, Octane will offer to download and install the RoadRunner binary for you.

RoadRunner Via Laravel Sail

If you plan to develop your application using Laravel Sail, you should run the following commands to install Octane and RoadRunner:

./vendor/bin/sail up

./vendor/bin/sail composer require laravel/octane spiral/roadrunner

Next, you should start a Sail shell and use the rr executable to retrieve the latest Linux based build of the RoadRunner binary:

./vendor/bin/sail shell

# Within the Sail shell...
./vendor/bin/rr get-binary

After installing the RoadRunner binary, you may exit your Sail shell session. You will now need to adjust the supervisor.conf file used by Sail to keep your application running. To get started, execute the sail:publish Artisan command:

./vendor/bin/sail artisan sail:publish

Next, update the command directive of your application's docker/supervisord.conf file so that Sail serves your application using Octane instead of the PHP development server:

command=/usr/bin/php -d variables_order=EGPCS /var/www/html/artisan octane:start --server=roadrunner --host=0.0.0.0 --rpc-port=6001 --port=80

Finally, ensure the rr binary is executable and build your Sail images:

chmod +x ./rr

./vendor/bin/sail build --no-cache

Swoole

If you plan to use the Swoole application server to serve your Laravel Octane application, you must install the Swoole PHP extension. Typically, this can be done via PECL:

pecl install swoole

Open Swoole

If you want to use the Open Swoole application server to serve your Laravel Octane application, you must install the Open Swoole PHP extension. Typically, this can be done via PECL:

pecl install openswoole

Using Laravel Octane with Open Swoole grants the same functionality provided by Swoole, such as concurrent tasks, ticks, and intervals.

Swoole Via Laravel Sail

Warning Before serving an Octane application via Sail, ensure you have the latest version of Laravel Sail and execute ./vendor/bin/sail build --no-cache within your application's root directory.

Alternatively, you may develop your Swoole based Octane application using Laravel Sail, the official Docker based development environment for Laravel. Laravel Sail includes the Swoole extension by default. However, you will still need to adjust the supervisor.conf file used by Sail to keep your application running. To get started, execute the sail:publish Artisan command:

./vendor/bin/sail artisan sail:publish

Next, update the command directive of your application's docker/supervisord.conf file so that Sail serves your application using Octane instead of the PHP development server:

command=/usr/bin/php -d variables_order=EGPCS /var/www/html/artisan octane:start --server=swoole --host=0.0.0.0 --port=80

Finally, build your Sail images:

./vendor/bin/sail build --no-cache

Swoole Configuration

Swoole supports a few additional configuration options that you may add to your octane configuration file if necessary. Because they rarely need to be modified, these options are not included in the default configuration file:

'swoole' => [
    'options' => [
        'log_file' => storage_path('logs/swoole_http.log'),
        'package_max_length' => 10 * 1024 * 1024,
    ],
],

Serving Your Application

The Octane server can be started via the octane:start Artisan command. By default, this command will utilize the server specified by the server configuration option of your application's octane configuration file:

php artisan octane:start

By default, Octane will start the server on port 8000, so you may access your application in a web browser via http://localhost:8000.

Serving Your Application Via HTTPS

By default, applications running via Octane generate links prefixed with http://. The OCTANE_HTTPS environment variable, used within your application's config/octane.php configuration file, can be set to true when serving your application via HTTPS. When this configuration value is set to true, Octane will instruct Laravel to prefix all generated links with https://:

'https' => env('OCTANE_HTTPS', false),

Serving Your Application Via Nginx

Note If you aren't quite ready to manage your own server configuration or aren't comfortable configuring all of the various services needed to run a robust Laravel Octane application, check out Laravel Forge.

In production environments, you should serve your Octane application behind a traditional web server such as a Nginx or Apache. Doing so will allow the web server to serve your static assets such as images and stylesheets, as well as manage your SSL certificate termination.

In the Nginx configuration example below, Nginx will serve the site's static assets and proxy requests to the Octane server that is running on port 8000:

map $http_upgrade $connection_upgrade {
    default upgrade;
    ''      close;
}

server {
    listen 80;
    listen [::]:80;
    server_name domain.com;
    server_tokens off;
    root /home/forge/domain.com/public;

    index index.php;

    charset utf-8;

    location /index.php {
        try_files /not_exists @octane;
    }

    location / {
        try_files $uri $uri/ @octane;
    }

    location = /favicon.ico { access_log off; log_not_found off; }
    location = /robots.txt  { access_log off; log_not_found off; }

    access_log off;
    error_log  /var/log/nginx/domain.com-error.log error;

    error_page 404 /index.php;

    location @octane {
        set $suffix "";

        if ($uri = /index.php) {
            set $suffix ?$query_string;
        }

        proxy_http_version 1.1;
        proxy_set_header Host $http_host;
        proxy_set_header Scheme $scheme;
        proxy_set_header SERVER_PORT $server_port;
        proxy_set_header REMOTE_ADDR $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header Upgrade $http_upgrade;
        proxy_set_header Connection $connection_upgrade;

        proxy_pass http://127.0.0.1:8000$suffix;
    }
}

Watching For File Changes

Since your application is loaded in memory once when the Octane server starts, any changes to your application's files will not be reflected when you refresh your browser. For example, route definitions added to your routes/web.php file will not be reflected until the server is restarted. For convenience, you may use the --watch flag to instruct Octane to automatically restart the server on any file changes within your application:

php artisan octane:start --watch

Before using this feature, you should ensure that Node is installed within your local development environment. In addition, you should install the Chokidar file-watching library within your project:

npm install --save-dev chokidar

You may configure the directories and files that should be watched using the watch configuration option within your application's config/octane.php configuration file.

Specifying The Worker Count

By default, Octane will start an application request worker for each CPU core provided by your machine. These workers will then be used to serve incoming HTTP requests as they enter your application. You may manually specify how many workers you would like to start using the --workers option when invoking the octane:start command:

php artisan octane:start --workers=4

If you are using the Swoole application server, you may also specify how many "task workers" you wish to start:

php artisan octane:start --workers=4 --task-workers=6

Specifying The Max Request Count

To help prevent stray memory leaks, Octane gracefully restarts any worker once it has handled 500 requests. To adjust this number, you may use the --max-requests option:

php artisan octane:start --max-requests=250

Reloading The Workers

You may gracefully restart the Octane server's application workers using the octane:reload command. Typically, this should be done after deployment so that your newly deployed code is loaded into memory and is used to serve to subsequent requests:

php artisan octane:reload

Stopping The Server

You may stop the Octane server using the octane:stop Artisan command:

php artisan octane:stop

Checking The Server Status

You may check the current status of the Octane server using the octane:status Artisan command:

php artisan octane:status

Dependency Injection & Octane

Since Octane boots your application once and keeps it in memory while serving requests, there are a few caveats you should consider while building your application. For example, the register and boot methods of your application's service providers will only be executed once when the request worker initially boots. On subsequent requests, the same application instance will be reused.

In light of this, you should take special care when injecting the application service container or request into any object's constructor. By doing so, that object may have a stale version of the container or request on subsequent requests.

Octane will automatically handle resetting any first-party framework state between requests. However, Octane does not always know how to reset the global state created by your application. Therefore, you should be aware of how to build your application in a way that is Octane friendly. Below, we will discuss the most common situations that may cause problems while using Octane.

Container Injection

In general, you should avoid injecting the application service container or HTTP request instance into the constructors of other objects. For example, the following binding injects the entire application service container into an object that is bound as a singleton:

use App\Service;
use Illuminate\Contracts\Foundation\Application;

/**
 * Register any application services.
 */
public function register(): void
{
    $this->app->singleton(Service::class, function (Application $app) {
        return new Service($app);
    });
}

In this example, if the Service instance is resolved during the application boot process, the container will be injected into the service and that same container will be held by the Service instance on subsequent requests. This may not be a problem for your particular application; however, it can lead to the container unexpectedly missing bindings that were added later in the boot cycle or by a subsequent request.

As a work-around, you could either stop registering the binding as a singleton, or you could inject a container resolver closure into the service that always resolves the current container instance:

use App\Service;
use Illuminate\Container\Container;
use Illuminate\Contracts\Foundation\Application;

$this->app->bind(Service::class, function (Application $app) {
    return new Service($app);
});

$this->app->singleton(Service::class, function () {
    return new Service(fn () => Container::getInstance());
});

The global app helper and the Container::getInstance() method will always return the latest version of the application container.

Request Injection

In general, you should avoid injecting the application service container or HTTP request instance into the constructors of other objects. For example, the following binding injects the entire request instance into an object that is bound as a singleton:

use App\Service;
use Illuminate\Contracts\Foundation\Application;

/**
 * Register any application services.
 */
public function register(): void
{
    $this->app->singleton(Service::class, function (Application $app) {
        return new Service($app['request']);
    });
}

In this example, if the Service instance is resolved during the application boot process, the HTTP request will be injected into the service and that same request will be held by the Service instance on subsequent requests. Therefore, all headers, input, and query string data will be incorrect, as well as all other request data.

As a work-around, you could either stop registering the binding as a singleton, or you could inject a request resolver closure into the service that always resolves the current request instance. Or, the most recommended approach is simply to pass the specific request information your object needs to one of the object's methods at runtime:

use App\Service;
use Illuminate\Contracts\Foundation\Application;

$this->app->bind(Service::class, function (Application $app) {
    return new Service($app['request']);
});

$this->app->singleton(Service::class, function (Application $app) {
    return new Service(fn () => $app['request']);
});

// Or...

$service->method($request->input('name'));

The global request helper will always return the request the application is currently handling and is therefore safe to use within your application.

Warning It is acceptable to type-hint the Illuminate\Http\Request instance on your controller methods and route closures.

Configuration Repository Injection

In general, you should avoid injecting the configuration repository instance into the constructors of other objects. For example, the following binding injects the configuration repository into an object that is bound as a singleton:

use App\Service;
use Illuminate\Contracts\Foundation\Application;

/**
 * Register any application services.
 */
public function register(): void
{
    $this->app->singleton(Service::class, function (Application $app) {
        return new Service($app->make('config'));
    });
}

In this example, if the configuration values change between requests, that service will not have access to the new values because it's depending on the original repository instance.

As a work-around, you could either stop registering the binding as a singleton, or you could inject a configuration repository resolver closure to the class:

use App\Service;
use Illuminate\Container\Container;
use Illuminate\Contracts\Foundation\Application;

$this->app->bind(Service::class, function (Application $app) {
    return new Service($app->make('config'));
});

$this->app->singleton(Service::class, function () {
    return new Service(fn () => Container::getInstance()->make('config'));
});

The global config will always return the latest version of the configuration repository and is therefore safe to use within your application.

Managing Memory Leaks

Remember, Octane keeps your application in memory between requests; therefore, adding data to a statically maintained array will result in a memory leak. For example, the following controller has a memory leak since each request to the application will continue to add data to the static $data array:

use App\Service;
use Illuminate\Http\Request;
use Illuminate\Support\Str;

/**
 * Handle an incoming request.
 */
public function index(Request $request): array
{
    Service::$data[] = Str::random(10);

    return [
        // ...
    ];
}

While building your application, you should take special care to avoid creating these types of memory leaks. It is recommended that you monitor your application's memory usage during local development to ensure you are not introducing new memory leaks into your application.

Concurrent Tasks

Warning This feature requires Swoole.

When using Swoole, you may execute operations concurrently via light-weight background tasks. You may accomplish this using Octane's concurrently method. You may combine this method with PHP array destructuring to retrieve the results of each operation:

use App\User;
use App\Server;
use Laravel\Octane\Facades\Octane;

[$users, $servers] = Octane::concurrently([
    fn () => User::all(),
    fn () => Server::all(),
]);

Concurrent tasks processed by Octane utilize Swoole's "task workers", and execute within an entirely different process than the incoming request. The amount of workers available to process concurrent tasks is determined by the --task-workers directive on the octane:start command:

php artisan octane:start --workers=4 --task-workers=6

When invoking the concurrently method, you should not provide more than 1024 tasks due to limitations imposed by Swoole's task system.

Ticks & Intervals

Warning This feature requires Swoole.

When using Swoole, you may register "tick" operations that will be executed every specified number of seconds. You may register "tick" callbacks via the tick method. The first argument provided to the tick method should be a string that represents the name of the ticker. The second argument should be a callable that will be invoked at the specified interval.

In this example, we will register a closure to be invoked every 10 seconds. Typically, the tick method should be called within the boot method of one of your application's service providers:

Octane::tick('simple-ticker', fn () => ray('Ticking...'))
        ->seconds(10);

Using the immediate method, you may instruct Octane to immediately invoke the tick callback when the Octane server initially boots, and every N seconds thereafter:

Octane::tick('simple-ticker', fn () => ray('Ticking...'))
        ->seconds(10)
        ->immediate();

The Octane Cache

Warning This feature requires Swoole.

When using Swoole, you may leverage the Octane cache driver, which provides read and write speeds of up to 2 million operations per second. Therefore, this cache driver is an excellent choice for applications that need extreme read / write speeds from their caching layer.

This cache driver is powered by Swoole tables. All data stored in the cache is available to all workers on the server. However, the cached data will be flushed when the server is restarted:

Cache::store('octane')->put('framework', 'Laravel', 30);

Note The maximum number of entries allowed in the Octane cache may be defined in your application's octane configuration file.

Cache Intervals

In addition to the typical methods provided by Laravel's cache system, the Octane cache driver features interval based caches. These caches are automatically refreshed at the specified interval and should be registered within the boot method of one of your application's service providers. For example, the following cache will be refreshed every five seconds:

use Illuminate\Support\Str;

Cache::store('octane')->interval('random', function () {
    return Str::random(10);
}, seconds: 5);

Tables

Warning This feature requires Swoole.

When using Swoole, you may define and interact with your own arbitrary Swoole tables. Swoole tables provide extreme performance throughput and the data in these tables can be accessed by all workers on the server. However, the data within them will be lost when the server is restarted.

Tables should be defined within the tables configuration array of your application's octane configuration file. An example table that allows a maximum of 1000 rows is already configured for you. The maximum size of string columns may be configured by specifying the column size after the column type as seen below:

'tables' => [
    'example:1000' => [
        'name' => 'string:1000',
        'votes' => 'int',
    ],
],

To access a table, you may use the Octane::table method:

use Laravel\Octane\Facades\Octane;

Octane::table('example')->set('uuid', [
    'name' => 'Nuno Maduro',
    'votes' => 1000,
]);

return Octane::table('example')->get('uuid');

Warning The column types supported by Swoole tables are: string, int, and float.