Add performance test to GPT for filters on code language search
With the proposed addition of adding filters to advanced search to filter on what code language the search should be performed it may be beneficial to add a performance test to Gitlab Performance Tool suite of tests to ensure these filters work under stress. We'll have to determine if we're going to use aggregations (someone with more knowledge can provide context here) which would be performed by Elasticsearch (if I'm understanding that correctly).
For thought and discussion:
- How frequently should this test run (nightly as with our other performance and compatibility tests?)
- If Elasticsearch is doing the aggregations we'll need to evaluate the cost benefit of doing these tests
- I am currently uncertain of the files that are used to populate the GPT search space. There may be a mismatch between what kind of code languages are there, and what kind do we want there.