0.1.10 • Published 10 years ago

various-cluster v0.1.10

Weekly downloads
48
License
-
Repository
github
Last release
10 years ago

various-cluster

easily create a mulit-cluster environment with different worker-types.

if you need some messaging inside your worker-types, take a look at msghub

Why another Cluster-Module?

This is a thin wrapper around the node native cluster module. There are currently no cluster-module-wrappers out there, which will spawn different Worker-Types which are not necessarly some http-servers. also this module enables you to take full control of all shutdown-mechanisms.

Getting Started

$ npm install various-cluster

require it in your code, and use it:

var variouscluster = require('various-cluster');

variouscluster.init({
  title: 'title of your master process',
  count: 2, // fallback for workers with no count defined
  shutdownTimout: 10000, // fallback for workers with no shutdownTimout defined
  workers: [
    {
      title: 'some-worker',  // title of this worker process
      exec: 'lib/some-worker.js', // location of your worker to be executed
      count: 1, // number of process to be spawned
      shutdownTimeout: 1000, // how long thisw workertype has time to shutdown
      config: {} // add some worker-config right here
    },
    {
      title: 'another-worker',
      exec: 'lib/another-worker.js',
      count: 1,
      shutdownTimeout: 1000,
      config: {}
    }
  ]
});

Documentation

api-docs: (open doc/index.html in your browser)

Creation

create your environment with including various-cluster in your code and call the init-method with your cluster-config (see more about config below!)

var yourConfig = {...};
var variouscluster = require('various-cluster');
variouscluster.init(yourConfig);

Workerspecific Config and Initialization

you can also pass a workerspecific config to your workers, so you dont need to read the config manually in every process.

just follow this pattern:

// in your master pass the workerconfig
variouscluster.init({
  ...
  workers: [
    {
      config: {
        "someKey": "someValue"
      }
    }
  ]
});
// and in your worker create a constructor function receiving this config
function SomeWorker(config) {
  this.config = config;
}
// and add an init-method to your Worker. This will instantly be called from various-cluster if present!
SomeWorker.prototype.init = function () {
  console.log(this.config.someKey); // will print 'someValue'
}

Logging

various cluster will try to use a logger-instance which is appended to the current process. this logger instance needs a 'crit' and a 'notice' method. if no process.logger is present, it just uses node.js native util.log()

// in your master create a logger:
process.logger = {
  crit: function (msg) {
    console.error(msg);
  },
  notice: function (msg) {
    console.log(msg);
  }
};

Shutdown

in every worker-process you can append a shutdown-Listener if you want to cleanup something in your application before shutting down:

process.on('shutdown', function (done) {
  // clean up here and call the done-callback when finished
  done();
});

if the cleanup takes longer than in your config defined, the process will be forced to exit!

uncaughtExceptions

per default if in the master an uncaught-exception happens, the whole application will gracefully shutdown.

if a worker has an uncaughtException, it depends on your configuration. per default this worker is gracefully shutdown and is reforked. if you configure a workertype with the config-parameter 'shutdownAll: true', the whole application will gracefully shutdown.

CONFIG which you provide in the init-method

config.title

process.title of the master process, DEFAULT: 'various-cluster-master'

config.count

workercount which should be used (if no count is given for a single worker this one is used), DEFAULT: os.cpus().length

config.shutdownTimeout

shutdownTimeout in ms (if no shutdownTimeout is given for a single worker this one is used), DEFAULT: 10000 ...

config.workers

an array with all workers (see more below)

config.workersx.title

process.title of the worker process, DEFAULT: 'various-cluster-worker'

config.workersx.exec

the local module-file which should be executed, DEFAULT: undefined

config.workersx.count

workercount which should be used, DEFAULT: config.count

config.workersx.shutdownTimeout

shutdownTimeout in ms, DEFAULT: config.shutdownTimeout

config.workersx.shutdownAll

shutdown all workertypes and workers if this worker has an uncaught exception, DEFAULT: false

config.workersx.config

your workerspecific config DEFAULT: undefined

Examples

see the examples-directory!

Contributing

In lieu of a formal styleguide, take care to maintain the existing coding style. Lint your code using Grunt

Release History

  • 0.1.10 final working shutdown mechanism which respects all possibilties

  • 0.1.9 fix second restart of workers who will just restart

  • 0.1.8 log err.stack on exceptions

  • 0.1.7 fix shutdown in workertype

  • 0.1.6 rewrite shutdown mechanism

  • 0.1.5 allow worker count to be 0

  • 0.1.4 force exit when no shutdown-handler

  • 0.1.4 force exit when no shutdown-handler

  • 0.1.3 fix clean shutdown

  • 0.1.2 better exception handling + cleaner shutdown

  • 0.1.1 Added some keywords

  • 0.1.0 Initial Release

Contributors

  • Bastian "hereandnow" Behrens
  • Gleb Kotov

License

Copyright (c) 2013 Valiton GmbH Licensed under the MIT license.

0.1.10

10 years ago

0.1.9

10 years ago

0.1.8

11 years ago

0.1.7

11 years ago

0.1.6

11 years ago

0.1.5

11 years ago

0.1.4

11 years ago

0.1.3

11 years ago

0.1.2

11 years ago

0.1.1

11 years ago

0.1.0

11 years ago