run-time managment of a node cluster

npm install strong-cluster-control
79 downloads in the last day
307 downloads in the last week
1 581 downloads in the last month

Run-Time Management of a Node Cluster

Module for run-time management of a node cluster.

It is an extension of the node cluster module, not a replacement.

An alternative to integrating strong-cluster-control into an application using the API is to use strong-supervisor, a runner-based wrapper that requires no code changes.

For full documentation, see StrongLoop Suite Documentation - Clustering applications.

  • runs size workers (optionally), and monitors them for unexpected death
  • run-time control of cluster through command line and API
  • soft shutdown as well as hard termination of workers
  • throttles worker restart rate if they are exiting abnormally

It can be added to an existing application using the node cluster module without modifying how that application is currently starting up or using cluster, and still make use of additional features.

The controller module allows the cluster to be controlled through the

  • clusterctl command line, or
  • API calls on the module


npm install --save strong-cluster-control

The command line:

npm install -g strong-cluster-control
clusterctl --help


To instantiate cluster-control:

var cluster = require('cluster');
var control = require('strong-cluster-control');

// global setup here...

    size: control.CPUS
}).on('error', function(er) {

if(cluster.isWorker) {
    // do work here...

To control the cluster, assuming my-server is running in /apps/:

clusterctl --path /apps/my-server/clusterctl set-size 4
clusterctl --path /apps/my-server/clusterctl status
worker count: 4
worker id 0: { pid: 11454 }
worker id 1: { pid: 11471 }
worker id 2: { pid: 11473 }
worker id 3: { pid: 11475 }

For more in-depth examples, see the chat server example, and the in-source example.

clusterctl: Command Line Interface

The clusterctl command line utility can be used to control a cluster at run-time. It defaults to communicating over the clusterctl named socket in the current working directory, but an explicit path or port can be provided.

It provides the following commands:

  • status: reports the status of the cluster workers
  • set-size: set cluster size to some number of workers
  • disconnect: disconnect all workers
  • fork: fork one worker

disconnect and fork cause the cluster size to change, so new workers will probably be started or stopped to return the cluster to the set size. They are primarily for testing and development.

npm loves you