Setup Telegraf+InfluxDB+Grafana to Monitor Windows

Monitoring Windows with Grafana is pretty easy, but there are multiple systems that have to be set up to work together.

Prerequisites:

  • Grafana
  • InfluxDB

Main Steps:

  1. Create an InfluxDB database and users for Telegraf and Grafana
  2. Install Telegraf on Windows and configure it
  3. Setup a data source and dashboards in Grafana

It really sounds more daunting than it is

InfluxDB setup

We want to create an InfluxDB database,  create a user for telegraf to write data into influx, and a user for Grafana to read data out of influx.  From an SSH terminal, the commands will be

influx
CREATE DATABASE telegraf
CREATE USER telegraf WITH PASSWORD 'telegraf123'
CREATE USER grafana WITH PASSWORD 'grafana123'
GRANT WRITE ON telegraf TO telegraf 
GRANT READ ON telegraf TO grafana

Install Telegraf

You can go to https://portal.influxdata.com/downloads to get links download the client for different OS’s. Grab the link for windows, which at this time is https://dl.influxdata.com/telegraf/releases/telegraf-1.5.3_windows_amd64.zip and download that file using whatever method suits you best.   Extra the contents to a location you like,  I use c:\Program Files\telegraf\.  now you will need to modify the contents of telegraf.conf, I like to use Notepad++ but any text editor should be fine.

I like to modify the section called [global_tags] and put machine identifiers in there.

[global_tags]
 environment = "production"

You can add as many different tags under there as you would like, it takes some time to figure out what will be useful here.

When you have that completed, update the section for InfluxDB with the needed info. Make sure to update the IP and the passwords to the correct ones for your install.  Also if needed on the destination machine add the port 8086

# Configuration for influxdb server to send metrics to
[[outputs.influxdb]]
 urls = ["http://192.168.86.167:8086"] # required
 database = "telegraf" # required
 precision = "s"
 timeout = "5s"
 username = "telegraf"
 password = "telegraf123"

Now run a command prompt as administrator. Change to the directory where you have telegraf and its config file and run the following command to test your config:

C:\Program Files\telegraf>telegraf.exe --config telegraf.conf --test

The output should include a bunch of lines like the following:

 >win_perf_counters,instance=Intel[R]\ Ethernet\ Connection\ [2]\ I219-V,objectname=Network\ Interface,host=DESKTOP-MAIN Packets_Received_Errors=0 1522202369000000000
 > win_perf_counters,instance=Qualcomm\ Atheros\ QCA61x4A\ Wireless\ Network\ Adapter,objectname=Network\ Interface,host=DESKTOP-MAIN Packets_Received_Errors=0 1522202369000000000
 > win_perf_counters,instance=Teredo\ Tunneling\ Pseudo-Interface,objectname=Network\ Interface,host=DESKTOP-MAIN Packets_Received_Errors=0 1522202369000000000
 > win_perf_counters,objectname=Network\ Interface,host=DESKTOP-MAIN,instance=Intel[R]\ Ethernet\ Connection\ [2]\ I219-V Packets_Outbound_Discarded=0 1522202369000000000
 > win_perf_counters,objectname=Network\ Interface,host=DESKTOP-MAIN,instance=Qualcomm\ Atheros\ QCA61x4A\ Wireless\ Network\ Adapter Packets_Outbound_Discarded=0 1522202369000000000
 > win_perf_counters,instance=Teredo\ Tunneling\ Pseudo-Interface,objectname=Network\ Interface,host=DESKTOP-MAIN Packets_Outbound_Discarded=0 1522202369000000000
 > win_perf_counters,instance=Intel[R]\ Ethernet\ Connection\ [2]\ I219-V,objectname=Network\ Interface,host=DESKTOP-MAIN Packets_Outbound_Errors=0 1522202369000000000
 > win_perf_counters,instance=Qualcomm\ Atheros\ QCA61x4A\ Wireless\ Network\ Adapter,objectname=Network\ Interface,host=DESKTOP-MAIN Packets_Outbound_Errors=0 1522202369000000000
 > win_perf_counters,instance=Teredo\ Tunneling\ Pseudo-Interface,objectname=Network\ Interface,host=DESKTOP-MAIN Packets_Outbound_Errors=2 1522202369000000000

If it doesn’t, it should include error information that will help you determine what the issue is.  Once that works, you can then install telegraf as a service by running the following:

C:\Program Files\telegraf>telegraf.exe --service install

The service will not start automatically the first time however,  so then to start it run

net start telegraf

Now you should have telegraf collecting data from windows on a regular basis and dumping that data into InfluxDB, the only thing remaining is to graph it

Grafana Setup

In Grafana, set up a new data source.  It should look like the following:

teelgraf source

Once that is set up, then you can go create a dashboard and add a graph.  I created the following graph:

Windows CPU Graph

The query is:

SELECT mean("Percent_Processor_Time") FROM "win_cpu" WHERE ("host" = 'DESKTOP-MAIN' AND "instance" != '_Total') AND time >= now() - 5m GROUP BY time(500ms), "instance" fill(linear)

This basically tells InfluxDB to go get all of the win_cpu values where the host tag is set as “DESKTOP-MAIN” and the counter instance is not _Total.  For CPU values this means it gets the individual totals so that I can graph each CPU. Make that an equals instead and you’ll get just the overall CPU usage instead of the breakdown.

Then I group by tag(instance) which is how you get one line (or series) per CPU (performance counter).  After that, I use an alias by to make the name “CPU ” followed by the instance value.  If you don’t do that, you end up with some funky named series that just aren’t pretty to look at.   If anyone finds this interesting (or even if they don’t probably) I will make a post about how to use template variables to generate a whole dashboard of graphs for a whole set of hosts automagically.

This is a lot the first time you do it, maybe even the second.  But it really pays off and gives you some amazing ways to monitor computers and servers and pays off big in the end.

2 thoughts on “Setup Telegraf+InfluxDB+Grafana to Monitor Windows

Leave a comment