Rudder

https://github.com/Normation/cf-clerk/pull/67

Could you explain the whole scheme of resolution of that one?

AFAIU, the idea is:

- to provide on the ncf side something that set a variable to the current technique to report on,
- when we generate the bundle sequence, before each ncf technique, we set the variable to the name of the technique, so that the reporting is done correctly.

Is it that ?

Also, could clearly explain in which ticket each part of the logic is implemented ?

The Techniques generated by ncf-builder have a unique name, which is also the bundle name
Rudder generates a csv file for reporting, in the format

technique_name;;class_prefix_${key};;@@RUDDER_ID@@;;component name;;component key

So far, we used only the class_prefix_${key} as a key to extract data from the this csv file, to get the line to report. Problem: we may have several time the same class_prefix_${key}, if we use in different technique the same component with same value.

Since we know the technique_name (we are calling it in the bundlesequence), we can use it in the key to search the csv file

However, we don't want to pass the technique name down the chain of generic method, as it would be clunky, so we inject it in the bundle current_technique_report_info, by calling this bundle with the current ncf technique name
And then, we change the logger_rudder to retrieve this value and expose the proper report line

This ticket part is to change the bundlesequence, to call current_technique_report_info(technique_name) before calling tehcnique_name

Applied in changeset clerk:commit:4b2f07d02386bd6fd38ee27457080ce667e9c548.

This bug has been fixed in Rudder 2.10.10 and 2.11.7, which were these days.

• Announcement 2.10 2.11
• Changelog 2.10 2.11
• Download information: https://www.rudder-project.org/site/get-rudder/downloads/