Skip to content

Include SLIs from dedicated in to the error budgets for stage groups

The GitLab application running in Dedicated environments emits the same metrics, and we already include recording rules for alerting. We could theoretically calculate Error Budgets for stage groups from these as well. If we did, we could allow stage groups to see how the application is performing for these customer's use cases.

There are some difficulties with this idea:

  1. Dedicated instances are currently isolated from our main monitoring infrastructure (Thanos), we will need to find a way to safely get these metrics out to present to developers
  2. Traffic share of GitLab.com is going to hide anything from Dedicated, so we'll have to somehow separate this: I don't think we can get this into a single number
  3. Debugging problems will be very hard, as we can't provide access to logs or direct access to all metrics for these environments to all developers using error budgets for stage groups.