ELK stack - Using Collectd for System monitoring

27 Oct 2014

Context

ELK stack (ElasticSearch, Logstash and Kibana) is often used for applicative monitoring, based on applications logs. However it would be useful to also aggregate System metrics in Kibana dashboards. I will present you a solution to add system (CPU, Memory, disk, network) indicators into your ELK stack. I will also add Oracle monitoring using the Collectd Oracle plugin.

Collectd

Collectd is a daemon which polls at regular interval (10s by default) metrics where it is run. This data may be send to any external system (as long as a plugin is available), a Logstash daemon in our case.

Build and install Collectd :

In this post, I will show you how to build Collect in order to use the Oracle plugin.

Download and untar

First step is simple : grab the Collectd sources from the official web site : http://www.collectd.org/download.shtml.

Unzip the sources on your disk :

tar xzf collect-5.4.1.tar.gz

Set Oracle environment variables

2 environments variables must be set before building the Oracle plugin : ORACLE_HOME and LD_LIBRARY_PATH :

export ORACLE_HOME=/path/to/oracle/installation
export LD_LIBRARY_PATH=$ORACLE_HOME/lib

Launch configure step

Once these 2 environment variables are set, we can configure the Build process to build Collectd, the Oracle plugin and install them in a specific directory (/path/to/my/custom/installation):

cd collect-5.4.2.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
./configure --prefix=/path/to/my/custom/installation --enable-oracle

Check that Oracle module and folder are reported as yes by the Configure step :

...
Libraries :
...
oracle . . . . . yes
...
Modules :
...
oracle . . . . . yes
...

make all install

If previous check is ok, then we can build and install Collectd and its plugins :

make all install

Collectd and Oracle (plus others) plugins are now installed in /path/to/my/custom/installation path.

Configure Collectd

Collect configuration is done in the file collect.conf located in etc folder.

Let’s edit the file and configure it this way :

xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

Outputing data to Logstash is done by the network plugin : we specify the Logstash IP and port.

Launching Collectd

Collectd will run as a daemon.

The script is located in the sbin directory :

cd /path/to/my/custom/installation
cd sbin
./collect -C ../etc/collectd.conf

Logstash configuration

Logstash’s configuration to receive Collectd events is quite simple : a collectd output plugin exists.

Just add this input plugin to your Logstash configuration :

input {
    ...
    collectd {}
    ...
}

Relaunch Logstash and events should begin to appear in your ElasticSearch clsuter :)

It’s time for you to play with Kibana ! Have fun.

If you liked this post, you can share it with your followers or follow me on Twitter!

JpBriend