Http Server Plugin

The HTTP Server Plugin provides an asynchronous HTTP server and includes additional reload strategies.

Installation

$ composer require phpstreamserver/http-server

Example of usage

server.php

use Amp\Http\Server\HttpErrorException;
use Amp\Http\Server\Request;
use Amp\Http\Server\Response;
use PHPStreamServer\Core\Server;
use PHPStreamServer\Plugin\HttpServer\HttpServerPlugin;
use PHPStreamServer\Plugin\HttpServer\HttpServerProcess;
use PHPStreamServer\Plugin\HttpServer\ReloadStrategy\MaxRequestsReloadStrategy;

$server = new Server();

$server->addPlugin(
    new HttpServerPlugin(
        // HttpServerPlugin configuration
    ),
);

$server->addWorker(
    new HttpServerProcess(
        // HttpServerProcess configuration
        name: 'Web Server',
        count: 2,
        listen: '0.0.0.0:8080',
        onRequest: function (Request $request, HttpServerProcess $worker): Response {
            return match ($request->getUri()->getPath()) {
                '/' => new Response(body: 'Hello world'),
                '/ping' => new Response(body: 'pong'),
                default => throw new HttpErrorException(404),
            };
        },
        reloadStrategies: [
            new MaxRequestsReloadStrategy(100),
        ],
    ),
);

exit($server->run());

Configuration

🔌 HttpServerPlugin

Option	Type	Default	Description
http2Enable	bool	true	Optional. Enables support for HTTP/2 protocol.
httpConnectionTimeout	int	60	Optional. Timeout duration for idle HTTP connections.
httpHeaderSizeLimit	int	32768	Optional. Maximum allowed size for HTTP headers.
httpBodySizeLimit	int	131072	Optional. Maximum allowed size for the HTTP request body.
gzipMinLength	int	860	Optional. Minimum response size required to enable gzip compression.
gzipTypesRegex	string	*	Optional. Regular expression to match content types eligible for gzip compression.

* #^(?:text/.*+|[^/]*+/xml|[^+]*\+xml|application/(?:json|(?:x-)?javascript))$#i

⚙️ HttpServerProcess

This worker type is designed to handle incoming HTTP requests asynchronously.

Option	Type	Default	Description
listen	string|Listen|Listen[]	not set	The address at which the server is listening.
name	string	"HTTP Server"	Optional. The name associated with the worker process.
count	int	1	Optional. The number of processes to start.
reloadable	bool	true	Optional. Whether the worker can be reloaded with the reload command.
user	string	not set	Optional. Unix user of process. Current user by default.
group	string	not set	Optional. Unix group of process. Current group by default.
onStart	Closure	not set	Optional. A callback function executed when the worker starts.
onRequest	Closure	not set	Optional. A callback function executed when an HTTP request is received.
onStop	Closure	not set	Optional. A callback function executed when the worker stops.
onReload	Closure	not set	Optional. A callback function executed when the worker reloads.
middleware	Middleware[]	not set	Optional. A list of middlewares for processing HTTP requests.
reloadStrategies	ReloadStrategy[]	not set	Optional. The strategies used to reload the worker.
serverDir	string	not set	Optional. The directory to serve static files from.
accessLog	bool	true	Optional. Whether to log incoming HTTP requests.
gzip	bool	false	Optional. Enables gzip compression.
connectionLimit	int	not set	Optional. The maximum number of connections per worker.
connectionLimitPerIp	int	not set	Optional. The maximum number of connections allowed per IP.
concurrencyLimit	int	not set	Optional. The maximum number of concurrent HTTP requests per worker.

🔄️ EachRequestReloadStrategy

This strategy reloads the worker after every HTTP request, making it primarily useful for debugging purposes.

🔄️ MaxRequestsReloadStrategy

This strategy reloads the worker after handling a specified number of HTTP requests.

Option	Type	Default	Description
maxRequests	int	not set	The maximum number of requests a worker can handle before being reloaded.
dispersionPercentage	int	0	Optional. Variability percantage to maxRequests. *

* Adds variability to the reload threshold to avoid restarting all workers simultaneously. For example, with maxRequests = 1000 and dispersionPercentage = 20, workers will be reloaded after handling between 800 and 1000 requests.