Setting up collectd plugins with Graphite and Grafana

Intro

Installing collectd could be trivial, although setting up monitoring for continuous time-series metric collection should be simpler. This post is aimed at helping sysadmins setup collectd and connect it to a graphite instance, so that all those metrics could later be viewed from Grafana instance.

Note:

This post is a spin off from the main satperf project. To take a look at how satperf works, refer to this post

The process

Step 1

Install collectd for your system, Install Graphite server elsewhere (recommended: separate machine).

---

Step 2

When that’s installed, take a look at your /etc/collectd.conf and add plugins from the list below, as suitable

# Allow collectd to log
LoadPlugin syslog

# Loaded Plugins:
LoadPlugin write_graphite
LoadPlugin cpu
LoadPlugin df
LoadPlugin disk
LoadPlugin interface
LoadPlugin irq
LoadPlugin load
LoadPlugin memory
LoadPlugin numa
LoadPlugin processes
LoadPlugin postgresql
LoadPlugin swap
LoadPlugin turbostat
LoadPlugin unixsock
LoadPlugin uptime

Note:

• If you’re installing this on Red Hat Satellite, you might wanna make additional changes as per this template in satperf

• For others, you might wanna take a look at the above mentioned link anyway, for it serves as a generic reference for /etc/collectd.conf

• Replace variable names in above referenced satperf’s collectd.conf as per following defaults / or change them as suitable:

Gotchas (for both Satellite setup as well as normal setup):

1. 10 refers to 10 seconds

2. end results on graphite-web UI show under Metrics:

{{graphite_prefix}}.{{inventory_hostname}}.<metrics as per output from plugins>

3. For candlepin password:

$ grep "jpa.config.hibernate.connection.password" /etc/candlepin/candlepin.conf | awk -F'=' '{print $2}'`

4. For Satellite Foreman password:

$ grep "password" /etc/foreman/database.yml | awk '{print $2}' | tr -d '"'

---

Step 3

Once this is done, reload collectd server and check logs on graphite server to make sure you’re able to receive data

Other Gotchas:

• Use iptables -F if unable to send collectd metrics
• To graph the other parameters in Grafana, this satellite generic dashboard file
  • To simply look at which metrics are touched in above dashboard template, refer to this satperf vars file
• FYI, I’ve got an open PR to add collectd info to sos report, as a plugin
• I’ll add some screenshots to this post ASAP.