Fluentd is a fully free and open-source log management tool designed for processing data streams.

In this presentation, I will explain the internal implementation details of Fluentd such as:

The bootstrap sequence of Fluentd, and how Fluentd loads plugins
How Fluentd parses the configuration file
How an input plugin passes the data to output plugins internally
How a BufferedOutput plugin processes data
The event-driven programming using cool.io, and effects of GVL
I will describe the behavior together with the knowledge obtained in the operation at production. While such detailed knowledge is not essential in the use of Fluentd, it should provide you a variety of useful findings for your operation.

DISCLAIMER: In this presentation, I do not explain, what is Fluentd, and how to use Fluentd. Audiences who have following experiences are desirable:

Have ever used Fluentd
Have ever written Fluentd plugin(s)