Skip to content

Latest commit

 

History

History
186 lines (140 loc) · 5.6 KB

README.md

File metadata and controls

186 lines (140 loc) · 5.6 KB
Raw

Sematext Agent Framework for Node.js

A framework for monitoring agents written in Node.js for sending metrics to Sematext Cloud or InfluxDB. The Sematext Node.js monitoring agent is built on top of this framework.

Functionality

  • Sender interface to Sematext backend receivers
  • Buffering metrics in case of network outages
  • Reconnect after failures
  • Logging functions
  • Configuration handling
  • Pluggable agents

Example to Implement a Monitoring Agent

Sematext Cloud supports the Influx Line Protocol for metric ingestion.

Agent modules must provide a start and stopfunction.

The function agent.addMetrics(metric) collects a metric, which is shipped in bulk requests via Influx Line Protocol to the metrics receiver.

All metric objects, must have the mandatory fields measurement, tags and fields:

  • measurement - the name of the measurement e.g. 'process.memry'
  • tags - key/value pairs, useful for filtering or aggregation of data
  • fields - numeric fields with the measurement values
const SpmAgent = require('spm-agent')
const client = new SpmAgent()
const os = require('os')

// configure client, with non-default values
// or use the file ./.spmagentrc in YAML format
// the default configuration contains values for Sematext Cloud US
// SpmAgent.Config.tokens.spm = process.env.MONITORING_TOKEN
SpmAgent.Config.influx = {
  dbName: 'metrics',
  // change receiver to Sematext Cloud EU
  // default is spm-receiver.sematext.com for US region
  host: 'spm-receiver.eu.sematext.com',
  port: 443,
  protocol: 'https'
}

// a monitoring agent needs a start an stop function
class MemoryMonitor {
  start (agent) {
    this.agent = agent
    // initialize your metrics collector ...
    // Typically agents collect metrics periodically, every N seconds. The time between // two collection activities is the collectionInterval, specified in milliseconds.
    this.tid = setInterval(function () {
      const measurement = {
        // measurment namespace
        measurement: 'process.memory',
        timestap: new Date(),
        tags: {
          role: 'frontend',
          'os.host': os.hostname()
          // The monitoring token can be set as 'tags.token' value
          // Routing a metrics to a different monitoring app
          // requires setting the `token` tag
          // ,token = process.env.OTHER_APP_TOKEN
        },
        // metrics names and values
        fields: {
          rss: process.memoryUsage().rss
        }
      }
      // pass metrics to sender schedule
      agent.addMetrics(measurement)
    }, SpmAgent.Config.collectionInterval)
  }

  stop () {
    if (this.tid) {
      clearInterval(this.tid)
    }
  }
}

client.createAgent(new SpmAgent.Agent(new MemoryMonitor()))

// log collected metrics to console
// observe all metrics for testing/debuggin
client.on('metrics', console.log)
// observe a specific metric using the measurment name
client.on('process.memory', console.log)

Metric Tags

The agent is able to use environment variables as metrics tags. Define a list of environment variables to be used as metric tags.

Let's assume this is the enviroment of the monitored application:

> env
TERM_PROGRAM=Apple_Terminal
TERM=xterm-256color
SHELL=/bin/bash
TMPDIR=/var/folders/v6/bh2q1xsn27g4d2g54z2ylv100000gn/T/
TERM_PROGRAM_VERSION=404.1
TERM_SESSION_ID=F12E0DBD-BF40-466D-85EC-08E89EDC3440
USER=stefan
PWD=/Users/stefan
HOME=/Users/stefan
LOGNAME=stefan
LC_CTYPE=UTF-8
DISPLAY=/private/tmp/com.apple.launchd.J60ybGpban/org.macosforge.xquartz:0
_=/usr/bin/env

You want to see later the performance metrics aggregated or filtered for specific user. In this case the agent should collect the user name as tag. You can find the user name in the process environment variable USER. Let's also assume the workingdirectory of the application PWD is another relevant identifier for your monitored application.

To configure the agent collecting USER and PWD as tags you can specify a commaseparated list of environment variables for the the agent configuration:

# add the values of env. variables USER, PWD as tags to metrics
export MONITORING_TAGS_FROM_ENV="USER, PWD"

The generated metrics include then the environment variable name and its value as tags. The result in Influx Line Protocol (ILP) format, produced by the agent for metric called swap:

swap, USER=stefan,PWD=/Users/stefan out=0i,in=0i 1576765680000000000

Define static tags in key:value format

export MONITORING_TAGS_FROM_ENV="organisation:sematext, service:api"

The result in in ILP format:

swap, organisation=sematext,service=api out=0i,in=0i 1576765680000000000

Both types of tags can be mixed:

export MONITORING_TAGS_FROM_ENV="organisation:sematext, service:api, USER, PWD"

The result in in ILP format:

swap, organisation=sematext,service=api,USER=stefan,PWD=/Users/stefan,os.host=imac.local out=0i,in=0i 1576765680000000000

The config file entry influx.tagsFromEnv in .spmagenrc works as well:

tokens: 
  spm: 'YOUR_MONITORING_TOKEN'
  infra: 'YOUR_INFRA_MONITROING_TOKEN'
influx:
  tagsFromEnv: 'organisation:sematext, USER, PWD' 
  dbName: 'metrics'
  host: spm-receiver.sematext.com
  protocol: https
  port: 443

Contribute

Let us know about monitoring agents you need, maybe you like to contribute with your domain expertise!

Related Modules