Description

In this part we'll highlight the relevance of having good quality logs that are easy to understand by both humans and machines. We'll describe the properties good logs must have (relevant contents, unambiguous timestamping, easy classification, integrity...), and the existing tools designed to meet these requirements, with examples for both developers (what and how to log) and system administrators (what to do with those logs). Real-world examples of good and bad practices will be shown, together with some ideas for improvement.

Description

This lecture introduces Systemtap, a free software (GPL) infrastructure to simplify the gathering of information about the running Linux system.

The lecture's first part points out in which situations administrators and developers are often lacking information when diagnosing performance or functional problems.

A short overview about the Linux Kernel is given, enabling the listener to get a feeling, which information is accessible and might be interesting in his specific use-case.

Next, a brief overview of Systemtap's inner workings is given.  Therefore, common probing techniques, their prerequisites, possible risks and downsides are presented. This part ends by showing how these techniques are accessible in Systemtap and what is needed to get them to run.

The lecture's last part introduces Systemtap's scripting language. It explains a couple of commonly used scripts and shows how the listener can practically introduce them to his own logging and monitoring environment.