socketnaut

Socketnaut

Scalable multithreaded Node.js servers made easy.

Introduction

Socketnaut makes scaling native Node.js servers easy. A Socketnaut Service typically consists of a TCP proxy and a pool of HTTP servers. Socketnaut will uniformly distribute incoming TCP sockets across the pool of allocated servers. This strategy allows for both distribution and parallel processing of incoming requests. Socketnaut consumes native Node.js servers (e.g., http.Server, https.Server, net.Server, tls.Server); hence, if you know the Node API, you already know how to build applications on Socketnaut!

Socketnaut can be used in order to scale the main module of web applications built on performant Node.js web frameworks (e.g., Fastify, Koa, Express). Please see the Examples section for instructions on how to do this.

Features

• Socketnaut requires 0 out-of-org dependencies. Socketnaut's dependencies are published and maintained by the FAR Analytics & Research org:
  • Streams Logger; a performant logger built from Node.js streams.
  • Port Agent; an RPC-like facility for making inter-thread function calls.
• The Socketnaut ServiceProxy and ServiceAgent constructors consume native Node net.Server, http.Server, https.Server, and tls.Server instances; you can configure them however you choose.
• Import Socketnaut as a Node.js module (see the Hello, World! example) or take advantage of the packaged type definitions and import it into your TypeScript project.

Table of contents

• Installation
• Concepts
• Usage
• Examples
• API
• Tuning strategies
• Client-Proxy socket remote address and port
• Logging
• Versioning
• Test
• Support

Installation

npm install socketnaut

Concepts

A Socketnaut Service consists of a ServiceProxy and a ServiceAgent.

Service Proxy

A ServiceProxy is used in order to bind a TCP server to a specified port (usu. a public interface). A ServiceProxy can be instantiated using the createServiceProxy function. The ServiceProxy uniformly distributes TCP connections to servers (e.g., HTTP servers) in the worker thread pool. The ServiceProxy manages the thread pool according to the values specified for the minWorkers and maxWorkers parameters or workerCount parameter.

Service Agent

A ServiceAgent coordinates the state of its server (e.g., the server's address) with its respective proxy. A ServiceAgent can be instantiated using the createServiceAgent function. It consumes a native Node.js server (e.g., net.Server, http.Server, https.Server, tls.Server). The Node.js server provided to the ServiceAgent may be used the same way it is used natively; hence, Socketnaut works with many popular Node.js web frameworks. Please see the Examples section for instructions on how to use Socketnaut with native Node.js servers and web application frameworks.

Usage

This simple "Hello, World!" app is a complete Socketnaut Service. You're looking at an ordinary Node.js web app, except that a ServiceProxy instance is created in the index.js module and a ServiceAgent instance is created in the scaled http_server.js module - that is all it takes to scale this web app. Scaling sophisticated web applications is just as easy.

Please see the Hello, World! example for a working implementation.

Implement the index.js module

Import the dependencies and create a Service Proxy.

This is the main thread.

import * as net from "node:net";
import { createServiceProxy } from "socketnaut";

const server = net.createServer(); // Configure this TCP server however you choose.

server.listen({ port: 3080, host: "0.0.0.0" });

const proxy = createServiceProxy({
  server,
  workerCount: 42, // Start 42 instances of the `http_server.js` module.
  workerURL: "./http_server.js", // The scaled module.
});

Implement the http_server.js module

Import the dependencies, implement the application, and create a Service Agent.

This is the scaled application.

import * as http from "node:http";
import { createServiceAgent } from "socketnaut";

const server = http.createServer(); // Configure this HTTP Server however you choose.

server.on("request", (req, res) => {
  for (let now = Date.now(), then = now + 100; now < then; now = Date.now()); // Block for 100 milliseconds.
  res.end("Hello, World!");
});

server.listen({ port: 0, host: "127.0.0.1" });
// Specifying port 0 here will cause the Server to listen on a random port.
// The Socketnaut Agent will communicate the randomly selected port to the ServiceProxy.

const agent = createServiceAgent({ server });

Examples

An instance of Hello, World! ^</Node.js>

Please see the Hello, World! example for a working implementation.

Use Socketnaut to scale a Fastify web application ^{</TypeScript>}

Please see the Fastify example for a working implementation.

Use Socketnaut to scale a Koa web application ^{</TypeScript>}

Please see the Koa example for a working implementation.

Use Socketnaut to scale an Express web application ^{</TypeScript>}

Please see the Express example for a working implementation.

Redirect HTTP connections to an HTTPS server ^{</TypeScript>}

Please see the Redirect HTTP to HTTPS example for a working implementation.

A TLS proxy and an HTTP redirect ^{</TypeScript>}

In the previous example, the TLS endpoint was in the worker thread; however, it doesn't need to be. Alternatively, TLS can be handled by the proxy server. Please see the A TLS Proxy and an HTTP Redirect example for a working implementation.

API

The ServiceProxy class

socketnaut.createServiceProxy(options)

• options <ServiceProxyOptions>

  • maxWorkers <number> Optional argument that specifies the maximum number of worker threads permitted.

  • minWorkers <number> Optional argument that specifies the minimum number of worker threads permitted. Default: 0

  • server <node:net.Server> or <node:tls.Server> A net.Server configured however you choose.

  • workersCheckingInterval <number> Optional argument that specifies the approximate interval (milliseconds) at which inactive worker threads will be cleaned up.

  • workerOptions <node:worker_threads.WorkerOptions> Optional WorkerOptions passed to the worker_threads.Worker constructor.

  • workerCount <number> Optional argument that specifies the number of worker threads to be spawned when Socketnaut starts. This setting will override minWorkers and maxWorkers.

  • workerURL <string> or <URL> The URL or path to the .js module file that contains the ServiceAgent instance. This is the module that will be scaled according to the values specified for minWorkers and maxWorkers. Please see the Examples section for how to specify the proxy's ServiceAgent module.

Returns: <socketnaut.ServiceProxy>

Creates a ServiceProxy. Each process may contain any number of ServiceProxys. However, all ServiceProxys run in the main thread; hence, the number of instances created in each process should be considered carefully.

Event: 'ready' The 'ready' event is emitted when the ServiceProxy has spawned its worker threads.

public serviceProxy.shutdown()

Returns: <Promise<Array<PromiseSettledResult<unknown>>>>

Performs a graceful shutdown. The Server is closed. Event listeners are removed. Worker threads are terminated asynchronously. The process does a clean exit (this assumes there aren't any remaining refs). The method returns a Promise that will resolve to an Array of PromiseSettledResult, where each element reflects the exit status of each worker thread. It will throw an Error if the Server is closed prior to being opened.

The ServiceAgent class

socketnaut.createServiceAgent(options)

• options <ServiceAgentOptions>

  • server <node:http.Server> or <node:https.Server> or <node:net.Server> or <node:tls.Server> A native Node.js Server configured however you choose.

Returns: <socketnaut.ServiceAgent>

Creates a ServiceAgent. Just one ServiceAgent may be instantiated for each worker; hence, this function will throw an Error if it is called more than once in a module.

public serviceAgent.requestProxySocketAddressInfo(socket)

• socket <net.Socket> The socket associated with the http.IncomingMessage i.e., http.IncomingMessage.socket.

Returns: <Promise<socketnaut.ProxySocketAddressInfo>>

The method returns a Promise that will resolve to an object that contains information that describes the proxy's socket tuple (i.e., in most cases this will contain the client's IP address and port).

Tuning strategies

Scaling can be tuned by specifying a minimum and maximum, or a specific number, of allocated worker threads to be spawned.

Relevant ServiceProxy constructor parameters

socketnaut.createServiceProxy(options)

• options <ServiceProxyOptions>

  • maxWorkers <number> Optional argument that specifies the maximum number of worker threads permitted.

  • minWorkers <number> Optional argument that specifies the minimum number of worker threads permitted. Default: 0

  • workersCheckingInterval <number> Optional argument that specifies the approximate interval (milliseconds) at which inactive worker threads will be cleaned up.

  • workerCount <number> Optional argument that specifies the number of worker threads to be spawned when Socketnaut starts. This setting will override minWorkers and maxWorkers.

The minWorkers argument specifies the minimum number of worker threads to be permitted in the thread pool. minWorkers worker threads will be instantiated when the Socketnaut proxy starts. Socketnaut will not allow the thread pool to drop below the specified threshold. However, if a worker thread throws an uncaught exception, Socketnaut will not attempt to automatically restart it, which could result in a thread pool below the specified threshold.

The maxWorkers argument is a hard limit on online threads; however, because thread termination is asynchronous it is possible for the combined count of online and liminal threads to briefly exceed this limit.

The workersCheckingInterval argument specifies the approximate interval at which Socketnaut will attempt to clean up inactive worker threads. If Socketnaut's proxy finds that a thread has 0 connections, Socketnaut will remove it from the pool and send it a notification requesting that it exit.

The workerCount argument sets the number of worker threads to be spawned when Socketnaut starts. When this argument is specified the effect is for both minWorkers and maxWorkers to be set to the value of workerCount.

By variously specifying minWorkers, maxWorkers, and workersCheckingInterval, or workerCount, you can tune Socketnaut according to the requirements of your environment.

Client-Proxy socket remote address and port

Socketnaut provides a facility for obtaining information about the client-proxy socket. When a proxied request is made to an http.Server, the request handler is passed a http.IncomingMessage. The remote address of the socket, accessed using http.IncomingMessage.socket.remoteAddress, will provide the remote address of the proxy (usu. 127.0.0.1) - not the remote address of the client. Implementations such as Proxy Protocol and the Forwarded HTTP header are commonly used in order to address this issue.

Socketnaut solves this problem by simply providing a MessageChannel facility for requesting information about the client-proxy socket. Call the ServiceAgent.requestProxySocketAddressInfo method with the request socket (e.g., req.socket) as an argument. The method returns a Promise that will resolve to a socketnaut.ProxySocketAddressInfo object that contains information that describes the proxy's socket tuple.

Example

const server = http.createServer();

const agent = createServiceAgent({ server });

server.on(
  "request",
  async (req: http.IncomingMessage, res: http.ServerResponse) => {
    const proxySocketAddressInfo = await agent.requestProxySocketAddressInfo(
      req.socket
    );
    console.log(proxySocketAddressInfo);
    /* Output
    {
        local: { address: '192.0.2.1', family: 'IPv4', port: 3443 },
        remote: { address: '198.51.100.1', family: 'IPv4', port: 35798 }
    }
    */
    res.end();
  }
);

The information returned by the ServiceAgent.requestProxySocketAddressInfo method can be used in order to associate the remote client address and port with each HTTP request e.g., for logging purposes.

Logging

By default Socketnaut logs to the console using the performant Streams Logger.

Changing the log level

You can set the log level on the Logger itself to a syslog logging level using the log.setLevel method. The default log Level is INFO.

Set the Service Proxy's log level to DEBUG.

index.js

import { Level } from 'socketnaut';
...
proxy.log.setLevel(Level.DEBUG);

Set the Service Agent's log level to DEBUG.

http_server.js

import { Level } from 'socketnaut';
...
agent.log.setLevel(Level.DEBUG);

Log to a file using a rotating file handler

Socketnaut's Logger may be configured however you choose. You can connect or disconnect Streams logging Nodes from the logging graph. You can reference the Nodes by importing them from Socketnaut's index. In this example Socketnaut's Logger, Formatter, and ConsoleHandler are imported.

import { logger, formatter, consoleHandler } from "socketnaut";

Once you have imported Socketnaut's logging Nodes, you can manipulate the logging graph as you choose using the Streams Logger API. You could, for example, configure the Service Proxy to log to a file instead of the console.

In this example a RotatingFileHandler is instantiated that will log messages to a file named socketnaut.log. Socketnaut's Formatter is disconnected from the ConsoleHandler and connected to the RotatingFileHandler.

import { formatter, consoleHandler, Level } from "socketnaut";
import { RotatingFileHandler } from "streams-logger";

const rotatingFileHandler = new RotatingFileHandler({
  path: "./socketnaut.log",
  level: Level.DEBUG,
});

formatter.disconnect(consoleHandler).connect(rotatingFileHandler);

You can use Socketnaut's logger instance for your logging purposes or use a logger of your choice. Please see the detailed Streams Logger documentation for further instructions on how to configure a Streams logging graph.

Versioning

The Socketnaut package adheres to semantic versioning. Breaking changes to the public API will result in a turn of the major. Minor and patch changes will always be backward compatible.

Excerpted from Semantic Versioning 2.0.0:

Given a version number MAJOR.MINOR.PATCH, increment the:

1. MAJOR version when you make incompatible API changes
2. MINOR version when you add functionality in a backward compatible manner
3. PATCH version when you make backward compatible bug fixes

Additional labels for pre-release and build metadata are available as extensions to the MAJOR.MINOR.PATCH format.

Test

Requirements

• A Controller will attempt to bind to port 3000.
• A Service Proxy for the HTTP Server will attempt to bind to port 3080.
• A Service Proxy for the HTTPS Server will attempt to bind to port 3443.
• The test script will use openssl in order to generate a certificate and key.

Run the test

Clone the repository.

git clone https://github.com/faranalytics/socketnaut.git

Change directory into the root of the repository.

cd socketnaut

Install dependencies.

npm install && npm update

Run the test.

npm test

Use the level=INFO argument to display the connection distribution.

npm test level=INFO

Support

If you have a feature request or run into any issues, feel free to submit an issue or start a discussion. You’re also welcome to reach out directly to one of the authors.

• Adam Patterson