Spike: Internal Event Tracking - improve unique to allow arbitrary properties

Problem

When instrumenting a metric I'd like to be able to create metric that are not only unique based on the attributes of our StandardContext (e.g. user_id, project_id) but any arbitrary property that I add to the event.

Example where unique merge_request_ids are counted in RedisHLL. We currently have no way to mirror this metric with Snowplow.

This issue cover the first spike on using self-describing events.

Desired Outcome

An MR that is using self-describing Snowplow events to let us track events that besides the standard context hold a merge request id.

The MR isn't expected to be merged, but should serve as a baseline for evaluating if we want to continue with self-describing events and which tasks we would have to complete in order to use self-describing events.

Possible Solution

...

Edited Aug 15, 2023 by Jonas Larsen
Assignee Loading
Time tracking Loading