Monitor the evolution of the usage of runner registration tokens on .com

In order to make an informed decision regarding the removal of runner registration tokens, we need to be able to follow the evolution of the usage of runner registration tokens on .com, namely for group runner managers and project runner managers. One way to do this is to use Internal Event Tracking tracking project and namespace identifiers. We should also track the registration method in additional_properties.property.

We can then define the following metrics:

• Total runner managers created

• Total runner managers created through runner registration token (filtering on additional_properties.property == registration_token):

    filter:
      property: "registration_token"

As of Feb 4th, 2025, we see that runners created in the last 7 days with the legacy runner registration method represent sightly over 2/3 of the runner managers contacting the GitLab instance:

gitlabhq_dblab> SELECT registration_type, "ci_runners".runner_type, COUNT(*) FROM "ci_runners" WHERE "ci_runners"."created_at" > (NOW() - interval '7 days') AND ci_runners.contacted_at > (NOW() - interval '7 days') GROUP BY registration_type, "ci_runners".runner_type
+-------------------+-------------+-------+
| registration_type | runner_type | count |
|-------------------+-------------+-------|
| 0 (reg token)     | 2  (group)  | 8598  |
| 0                 | 3 (project) | 4196  |
| 1 (auth token)    | 2           | 2411  |
| 1                 | 3           | 2739  |
+-------------------+-------------+-------+
SELECT 4

Checking ci_finished_builds on ClickHouse for CI jobs executed by those runners, we get roughly the same ratios:

jobs executed by runners	count
reg token	728351
auth token	305519