General Info node.d

node.js plugins

Node.js is perfect for asynchronous operations. It is very fast and quite common (actually the whole web is based on it). Since data collection is not a CPU intensive task, node.js is an ideal solution for it.

node.d.plugin is a netdata plugin that provides an abstraction layer to allow easy and quick development of data collectors in node.js. It also manages all its data collectors (placed in /usr/libexec/netdata/node.d) using a single instance of node, thus lowering the memory footprint of data collection.

Of course, there can be independent plugins written in node.js (placed in /usr/libexec/netdata/plugins). These will have to be developed using the guidelines of External Plugins.

To run node.js plugins you need to have node installed in your system.

In some older systems, the package named node is not node.js. It is a terminal emulation program called ax25-node. In this case the node.js package may be referred as nodejs. Once you install nodejs, we suggest to link /usr/bin/nodejs to /usr/bin/node, so that typing node in your terminal, opens node.js. For more information check the Installation guide.

configuring node.d.plugin

node.d.plugin can work even without any configuration. Its default configuration file is /etc/netdata/node.d.conf (to edit it on your system run /etc/netdata/edit-config node.d.conf).

debugging modules written for node.d.plugin

To test node.d.plugin collectors, which are placed in /usr/libexec/netdata/node.d, you can run node.d.plugin by hand, with something like this:

/usr/libexec/netdata/plugins.d/node.d.plugin debug 1 X Y Z

node.d.plugin will run in debug mode (lots of debug info), with an update frequency of 1 second, evaluating only the collector scripts X (i.e. /usr/libexec/netdata/node.d/X.node.js), Y and Z. You can define zero or more modules. If none is defined, node.d.plugin will evaluate all modules available.

Keep in mind that if your configs are not in /etc/netdata, you should do the following before running node.d.plugin:

export NETDATA_CONFIG_DIR="/path/to/etc/netdata"

---

developing node.d.plugin modules

Your data collection module should be split in 3 parts:

• a function to fetch the data from its source. node.d.plugin already can fetch data from web sources, so you don't need to do anything about it for http.

• a function to process the fetched/manipulate the data fetched. This function will make a number of calls to create charts and dimensions and pass the collected values to netdata. This is the only function you need to write for collecting http JSON data.

• a configure and an update function, which take care of your module configuration and data refresh respectively. You can use the supplied ones.

Your module will automatically be able to process any number of servers, with different settings (even different data collection frequencies). You will write just the work needed for one and node.d.plugin will do the rest. For each server you are going to fetch data from, you will have to create a service (more later).

writing the data collection module

To provide a module called mymodule, you have create the file /usr/libexec/netdata/node.d/mymodule.node.js, with this structure:

// the processor is needed only
// if you need a custom processor
// other than http
netdata.processors.myprocessor = {
	name: 'myprocessor',

	process: function(service, callback) {

		/* do data collection here */

		callback(data);
	}
};

// this is the mymodule definition
var mymodule = {
	processResponse: function(service, data) {

		/* send information to the netdata server here */

	},

	configure: function(config) {
		var eligible_services = 0;

		if(typeof(config.servers) === 'undefined' || config.servers.length === 0) {

			/*
			 * create a service using internal defaults;
			 * this is used for auto-detecting the settings
			 * if possible
			 */

			netdata.service({
				name: 'a name for this service',
				update_every: this.update_every,
				module: this,
				processor: netdata.processors.myprocessor,
				// any other information your processor needs
			}).execute(this.processResponse);

			eligible_services++;
		}
		else {

			/*
			 * create a service for each server in the
			 * configuration file
			 */

			var len = config.servers.length;
			while(len--) {
				var server = config.servers[len];

				netdata.service({
					name: server.name,
					update_every: server.update_every,
					module: this,
					processor: netdata.processors.myprocessor,
					// any other information your processor needs
				}).execute(this.processResponse);

				eligible_services++;
			}
		}

		return eligible_services;
	},

	update: function(service, callback) {
		
		/*
		 * this function is called when each service
		 * created by the configure function, needs to
		 * collect updated values.
		 *
		 * You normally will not need to change it.
		 */

		service.execute(function(service, data) {
			mymodule.processResponse(service, data);
			callback();
		});
	},
};

module.exports = mymodule;

configure(config)

configure(config) is called just once, when node.d.plugin starts. The config file will contain the contents of /etc/netdata/node.d/mymodule.conf. This file should have the following format:

{
	"enable_autodetect": false,
	"update_every": 5,
	"servers": [ { /* server 1 */ }, { /* server 2 */ } ]
}

If the config file /etc/netdata/node.d/mymodule.conf does not give a enable_autodetect or update_every, these will be added by node.d.plugin. So you module will always have them.

The configuration file /etc/netdata/node.d/mymodule.conf may contain whatever else is needed for mymodule.

processResponse(data)

data may be null or whatever the processor specified in the service returned.

The service object defines a set of functions to allow you send information to the netdata core about:

1. Charts and dimension definitions
2. Updated values, from the collected values

---

FIXME: document an operational node.d.plugin data collector - the best example is the snmp collector