Include Sidekiq SLIs into error budget calculation
The Rails monolith contributes metrics to several services: web, api and git all have a puma component that receives metrics labelled with a feature category that defines the SLI for that component. In #945 (closed) we've added the recording rules to aggregate those up into metrics for stage groups that can be used to calculate an error budget.
Now we need to do the same for Sidekiq. To do that, we will need to extend the perWorkerRecordingRules to add gitlab_background_jobs:execution:apdex:success:rate_<interval>.
Then we can use these per-worker rules including a feature category label, to record the following feature category metrics metrics labelled with the sidekiq-execution component:
| per worker metric | feature category metrics |
|---|---|
| gitlab_background_jobs:execution:ops:rate_%s | gitlab:component:feature_category:execution:ops:rate_%s |
| gitlab_background_jobs:execution:error:rate_%s | gitlab:component:feature_category:execution:error:rate_%s |
| gitlab_background_jobs:execution:apdex:weight:score_%s | gitlab:component:feature_category:execution:apdex:weight:score_%s |
| gitlab_background_jobs:execution:apdex:success:rate_%s | gitlab:component:feature_category:execution:apdex:success:rate_%s |
When these metrics are recorded they will be included in the error budget calculation