Fluentbit: Processing multiline/stack trace log messages (java, python, go, .)

[ricsanfre]

Enhancement Request

In current implementation application stack traces (java, go, python), multi-line logs, are split into different log entries.

Activate fluentbit new built-in mutiline parsers/filters (availible since v1.8.2) to concatenate Multiline or Stack trace log messages. In my current implementation, multiline parser is configured for Tail input (using cri parsers) to parse possible multiline containerd logs.
It is needed to apply mutiline filter to the log field, discovered while parsing containerd log (cri parse). to process this stack traces generated by different languages (go, python, java).
See furthter details multiline filter doc.

Implementation details

Filter like the following should be added

    [FILTER]
        name                  multiline
        match                 *
        multiline.key_content log
        multiline.parser      java,python,go

[ricsanfre]

There is a interesting blog where multiline parsing in fluentd is showed. The github repository attached to the blog contains a testing app that can be used for testing purposes.
This test application is built using Spring boot framework.

The process to build such application, using docker, is defined here: how to build docker images for java