The goal of this project would be to enhanc the GraphQL SLI we've started in https://gitlab.com/gitlab-org/gitlab/-/issues/329288 so it can properly feed into the error budget for stage groups just like rails requests. Including an `apdex` and a `error` component. ## Proposal We were only measuring the apex-successes for GraphQL queries and use that SLI for our `api` service, we needed to allow stage groups to customize the urgency for queries. We also needed to extend the SLI so it includes error rates separately from the request, allowing us to pinpoint which queries issues by clients we own are causing errors and correlate that to groups. The GraphQL SLI has been created but only measures apdex. The default urgency currently has a 1s target duration, but this probably needs to be revisited. All stage groups have been opted out from using the SLI in their error budget by default Issues in `gitlab-org/gitlab`: - https://gitlab.com/gitlab-org/gitlab/-/issues/349546: Refine the urgencies defined for GraphQL queries - https://gitlab.com/gitlab-org/gitlab/-/issues/345263: Graphql query error SLI - https://gitlab.com/gitlab-org/gitlab/-/issues/328535: GraphQL query-to-feature correlation mechanism - https://gitlab.com/gitlab-org/gitlab/-/issues/345141: GraphQL query-to-urgency correlation