Overview

MariaDB Feedback plugin has a special set of status variables of the type:
FEATURE_CHECK_CONSTRAINTS, FEATURE_JSON, FEATURE_APPLICATION_TIME_PERIOD, etc.

We want to count how many servers make use of any of these features. A feature is used when the counter for that feature is greater than 0.

The end goal is to produce a line chart that shows feature-usage over time. We want a series for each FEATURE.

Required steps:

Create a data-extractor to extract FEATURE based data from raw upload data.

The upload-fact should store JSON values of the form:

ComputedUploadFact(key='features',

                   value=('{'

        '"json": true,'

        '"application_time_period": true,'

        '"window_functions": true,'

        '"fulltext": true'

    }'))

For simplicity, we'll only store a key as true if the feature is used, otherwise
it will be absent.

Create a multi-series chart for each feature

Implementation suggestion:

• In feedback_plugin/data_processing/charts.py ->
• Function to compute a "total number of distinct servers where feature xxx is used, grouped by month"
• Implementation idea: The feature name could be a parameter for this function.

Hints:

Most of the relevant code can be found in:

• feedback_plugin/data_processing/extractors.py
• feedback_plugin/data_processing/etl.py

• Check ServerVersionExtractor for an UploadFactExtractor example.
• To see how an UploadFactExtractor is called by the ETL framework, you can look at get_upload_data_for_data_extractors.