A wrapper to enable the use of any in-process store with node cluster via cluster process and worker messages (i.e. for Raspberry Pi servers).
Go to file
AJ ONeal 34b0aaf460 use cluster.on('fork') for adding forks 2016-09-08 18:29:12 -06:00
.gitignore v2.0.0 2016-09-08 18:00:53 -06:00
LICENSE v2.0.0 2016-09-08 18:00:53 -06:00
README.md use cluster.on('fork') for adding forks 2016-09-08 18:28:48 -06:00
index.js v2.0.0 2016-09-08 18:00:53 -06:00
master.js use cluster.on('fork') for adding forks 2016-09-08 18:28:48 -06:00
memstore.js v2.0.0 2016-09-08 18:00:53 -06:00
package.json v2.0.1 2016-09-08 18:09:54 -06:00
test.js use cluster.on('fork') for adding forks 2016-09-08 18:29:12 -06:00
worker.js v2.0.0 2016-09-08 18:00:53 -06:00

README.md

cluster-store

Makes any storage strategy similar to express/session useful in both cluster and non-cluster environments by wrapping it with cluster-rpc.

Also works with level-session-store (leveldb), connect-session-knex (SQLite3), session-file-store (fs), and any other embedded / in-process store.

Note: Most people would probably prefer to just use Redis rather than wrap a dumb memstore as a service... but I am not most people.

Install

npm install --save cluster-store@2.x

v1.x vs v2.x

The old v1 used ws which makes it usable when clustering node processes without using cluster.

If you need that functionaliy, use v1 instead of v2.

Usage

standalone (non-cluster)


The usage is exactly the same as master, no changes necessary.

master

In the master/standalone process you will create the real store instance.

If you need to manually specify which worker will be enabled for this funcitonality you must set addOnFork to false and call addWorker() manually.

'use strict';

var cluster = require('cluster');

var cstore = require('cluster-store/master').create({
  name: 'foo-store'
, store: null             // use default in-memory store
, addOnFork: true         // default
});

// if you addOnFork is set to false you can add specific forks manually
//cstore.addWorker(cluster.fork());

cstore.then(function (store) {
  store.set('foo', 'bar');
});

Note: store can be replaced with any express/session store, such as:

  • new require('express-session/session/memory')()
  • require('level-session-store')(session)
  • and others

worker

'use strict';

// retrieve the instance
var cstore = require('cluster-store/worker').create({
  name: 'foo-store'
});

cstore.then(function (store) {
  store.get('foo', function (err, result) {
    console.log(result);
  });
});

API

This is modeled after Express' Session Store Implementation

Note: These are only exposed if the underlying store supports them.

CRUD methods

  • store.set(id, data, fn) => (error)
  • store.get(id, fn) => (error, data)
  • store.touch(id, data, fn) => (error)
  • store.destroy(id, fn) => (error)

Helpers

  • store.all(fn) => (error, array)
  • store.clear(fn) => (error)
  • store.length(fn) => (error, length)

See https://github.com/expressjs/session#session-store-implementation@4.x for full details

Example

'use strict';

var cluster = require('cluster');
var cstore;


if (cluster.isMaster) {


  cstore = require('./master').create({
    name: 'foo-level'
  });
  cstore.then(function (store) {
    store.put('foo', 'bar');
  });

  // create as many workers as you need
  cluster.fork();


}
else {


  cstore = require('./worker').create({
    name: 'foo-level'
  });


}


cstore.then(function (store) {
  setTimeout(function () {
    store.get('foo', function (err, result) {
      console.log(cluster.isMaster && '0' || cluster.worker.id.toString(), "store.get('foo')", result);

      if (!cluster.isMaster) {
        process.exit(0);
      }
    });
  }, 250);
});

process.on('unhandledRejection', function (err) {
  console.log('unhandledRejection', err);
});