Monitoring nginx with CollectD,Prometheus and grafana

Nginx is a popular web server which is is used as HTTP server,Loadbalancer and Generic TCP/UDP proxy. It has been running on many heavily loaded webservers and powering popular websites like Netflix, and many more.  In this article we will explain monitoring nginx with CollectD, Prometheus and Grafana. Nginx monitoring is useful for ensuring your web server availability and performance remains high.

The basic Nginx monitoring stats are provided by HttpStubStatusModule – metrics include requests/sec and number of open connections, along with stats for how fast requests are being handled. Let us see how to collect this metrics using Collectd, store the metrics data in time series database Prometheus and visualize it using Grafana.

Following components are there in this monitoring stack

  • CollectD : Collect the metrics from nginx HttpStubStatusModule and forward to “Collectd Exporter” using “network” module
  •  Collectd Exporter : Accepts the metrics send by collectd ,transforms and exposes them for consumption by Prometheus.
  • Prometheus : Polls the HTTP endpoint exposed by “Collectd Exporter” , pull the metrics data and store it in time series format.
  • Grafana : Visualize the metrics data stored in Prometheus.

Now let us see how these components works together

1. Make sure HttpStubStatusModule is enabled in nginx

First we have to check if the “stub status” module is enabled in nginx. Use the following command to list the enabled modules and flags inside nginx

If not, you will have to recompile Nginx with stub module . Instructions are here

Configure stub module

Add the following directives inside your nginx “server” block

Once done with configuration changes, restart nginx on the machine. You can verify if the the module is working, by launching following url in browser or any shell client

2. Configure CollectD to get the metrics data from Nginx and forward it to “Collectd Exporter”

Add the following directives in the Collectd configuration

Restart collectd once done with configuration changes

3.  Collectd Exporter

Unlike other monitoring stacks, like Graphite, Prometheus pull metrics from servers by scraping metrics endpoints exposed on each servers. As Collectd doesn’t have such endpoint, we will install “collectd exporter”

Collectd Exporter for Prometheus accepts collectd’s binary network protocol as sent by collectd’s network plugin and transforms and exposes them for consumption by Prometheus.

Download “collectd_exporter”

extract it using following command

Now we can run it using

start collectd_exporter with “-collectd.listen-address=”:25826″ to start consuming and exporting the metrics metrics data sent by collectd.

4. Configure Prometheus

Finally we will configure Prometheus to Poll the “Collectd Exporter” for metrics data.

Download latest version of Prometheus binary from their github repo

There are two main files under this folder “prometheus-1.2.1.linux-amd64” , one is Prometheus binary and other one is config file prometheus.yml

Now we need to configure Prometheus to scrape the HTTP endpoint exposed by “collectd exporter” and collect the metrics

Add the following line in the prometheus.yml file

Config file will look like below

Now we will start Prometheus , by running following command

You can see it is started and listening on port 9090. Now we can access the Prometheus UI by launching the following URL in browser ( replace the IP address :))

If everything goes fine, you should be able to see the metrics inside Prometheus UI . Please see screenshot below



5. Visualize metrics data using Grafana

Now we can configure Grafana to display the data stored in Prometheus. I have created a video on this. please check it out

Feel free to contact us if you face any issue ????

Leave a Reply

Your email address will not be published. Required fields are marked *