Skip to content

Latest commit

 

History

History
208 lines (152 loc) · 6.28 KB

README.md

File metadata and controls

208 lines (152 loc) · 6.28 KB
Raw

go-custom-exporter

This package intends to provide a way to easily create Prometheus exporters from the output of any custom script/program without having to worry about doing much code.

Dependencies

For this package to be successfully built, the following dependencies are required:

Download

Download the binaries from the latest release to your local machine and place it in your PATH and follow the instructions below for usage.

Usage

In order to execute the script follow the example below:

$ prom_exporter -script <script_path> -port <port> -timeout <seconds>

For example:

$ prom_exporter -script /opt/my_script.sh -port 8080  -timeout 5

Setting-up Your Own Scripts

Before using prom_exporter to export your own custom scripts, first you have to be sure you have a script/program/utility ready to return data in the required format as shown below:

system, subsystem, metric_name, metric_value
system, subsystem, metric_name, metric_value
system, subsystem, metric_name, metric_value

(...)

Consider the simple bash script below as an example, which should print the values from /proc/loadavg in the required format:

#!/bin/bash

function printMetric {
    SYSTEM=$1
    SUBSYSTEM=$2
    METRIC_NAME=$3
    METRIC_VALUE=$4
    echo "$SYSTEM $SUBSYSTEM $METRIC_NAME $METRIC_VALUE"
}

while read -r load1min load5min load15min jobs lastpid; do
    printMetric "demo01," "instance_01," "loadscript_load1," "$load1min"
    printMetric "demo01," "instance_01," "loadscript_load5," "$load5min"
    printMetric "demo01," "instance_01," "loadscript_load15," "$load15min"
    while IFS='/' read -r running background; do
        printMetric "demo01," "instance_01," "loadscript_jobs_running," "$running"
        printMetric "demo01," "instance_01," "loadscript_jobs_background," "$background"
    done <<< "$jobs"
    printMetric "demo01," "instance_01," "loadscript_pid_last," "$lastpid"
done < /proc/loadavg

The following output following is expected:

$ /opt/procloadavg.sh
demo01, loadscript_load1, instance_01, 1.62
demo01, loadscript_load5, instance_01, 1.40
demo01, loadscript_load15, instance_01, 1.52
demo01, loadscript_jobs_running, instance_01, 6
demo01, loadscript_jobs_background, instance_01, 1472
demo01, loadscript_pid_last, instance_01, 24586

Now execute the script with prom_exporter so that its data is ready to be scraped by Prometheus, try the following:

$ prom_exporter -script /opt/procloadavg.sh -port 8080

Now when accessing the http://localhost:8080/metrics the following text should be presented as expected by Prometheus for the scraping to work:

$ curl -s localhost:8080/metrics | grep custom
# HELP prom_custom_exporter Prometheus Gauge Metrics from Custom script/command exporter
# TYPE prom_custom_exporter gauge
prom_custom_exporter{metric="loadscript_jobs_background",subsystem="instance_01",system="demo01"} 1498
prom_custom_exporter{metric="loadscript_jobs_running",subsystem="instance_01",system="demo01"} 9
prom_custom_exporter{metric="loadscript_load1",subsystem="instance_01",system="demo01"} 1.33
prom_custom_exporter{metric="loadscript_load15",subsystem="instance_01",system="demo01"} 1.29
prom_custom_exporter{metric="loadscript_load5",subsystem="instance_01",system="demo01"} 1.38
prom_custom_exporter{metric="loadscript_pid_last",subsystem="instance_01",system="demo01"} 15763

Running Your Exporter as a Service with Systemd

  1. Create the service file as shown below:
$ sudo touch /lib/systemd/system/prom_exporter_loadavg.service
  1. Edit the service file with the following content:
[Unit]
Description=loadavg Prometheus Exporter

[Service]
Type=simple
Restart=always
RestartSec=5s
ExecStart=/usr/bin/prom_exporter -script /opt/procloadavg.sh -port 8080

[Install]
WantedBy=multi-user.target
  1. Enable the new systemd service as shown:
$ sudo systemctl enable prom_exporter_loadavg.service
  1. Restart your newly created service:
$ systemctl restart  prom_exporter_loadavg

Prometheus Scraping Settings

Add the following to your prometheus.yml configuration file:

  - job_name: 'custom_exporter'
    scrape_interval: 5s
    static_configs:
      - targets: ['<name_of_your_host>:<port_number>']

Restart the Prometheus Server, and verify at the Prometheus target list if the new job is listed as healthy, as shown below:

You can also confirm that data is being collected at the query page by searching for prom_custom_exporter, as shown below:

In order to get the full instructions on how to configure Prometheus to Scrape Your Custom Exporters follow the steps described at the following link from Prometheus official documentation:

Building from Source

To build the binaries from source, execute the following (if you already have GOPATH configured on your local machine):

$ go get github.com/fsilveirgo-custom-exporter

Or clone the repository directly with the following command:

git clone [email protected]:fsilveir/go-custom-exporter.git

After succesfully downloading the files from the repository, execute the script build.sh, as shown below:

~/go/src/github.com/fsilveir/go-custom-exporter$ ./build.sh 
Revision is 1172888
Building GOOS=windows GOARCH=amd64...
Building GOOS=windows GOARCH=386...
Building GOOS=linux GOARCH=amd64...
Building GOOS=linux GOARCH=386...
Building GOOS=freebsd GOARCH=amd64...
Building GOOS=freebsd GOARCH=386...
Building GOOS=darwin GOARCH=amd64...
Building GOOS=darwin GOARCH=386...
prom_exporter_darwin_386: OK
prom_exporter_darwin_amd64: OK
prom_exporter_freebsd_386: OK
prom_exporter_freebsd_amd64: OK
prom_exporter_linux_386: OK
prom_exporter_linux_amd64: OK
prom_exporter_windows_386.exe: OK
prom_exporter_windows_amd64.exe: OK