collectd is a UNIX daemon which periodically collects, analyzes, and stores performance data of the systems its running on, networking and data-center equipment and various other devices. Those statistics may then be used to find performance bottlenecks (performance analysis) and predict future system load (capacity planing). The main design goal is flexibility in how the data is acquired and further processed. The main distribution ships more than 100 plugins to collect metrics like, for example, CPU and memory usage, network and disk IO, various temperature information, application specific information (e.g. database servers) or detailed networking information. Through the modular design, collectd can monitor anything — ranging from a WRT home-router to a Cisco 7600 and scaling from a desktop computer up to automatically managed cloud infrastructures or other setups with several hunderts or thousands of hosts.
This talk provides an overview over the daemon and a selection of typical plugins. Also, the network plugins and its manifold operation modes will be explained in detail as well as the daemons many possibilities to extend the software and combine it with third-party tools (like Nagios). Optional further topics : querying values using SNMP ; writing extensions in C, Perl, Python, Java, ... ; tuning RRDtool in order to handle large amounts of data.
Sebastian ’tokkee’ Harl is engaged in Linux and Open Source software for many years. Besides actively working on projects like Debian (www.debian.org), collectd (www.collectd.org) and RRDtool (oss.oetiker.ch/rrdtool), he actively supports various Open Source events (e.g., FrOSCon,www.froscon.org, Grazer Linuxtage, linuxtage.at) by staffing booths and giving talks. Since 2011, we works as Linux, monitoring, and performance analysis consultant and trainer at teamix GmbH (www.teamix.net) in Nuremberg, Germany.