Start tracking MTBF for deployment-related incidents
Delivery Team's main performance indicator is MTTP but without a counterbalance, we risk moving too fast and impacting customers. As Delivery work towards hitting the new 12-hour MTTP target I'd like us to track, and own, MTBF for deployment-related incidents.
We already create incidents for anything deployment-related including:
- Pipeline failures, including failing QA tests, config issues, or issues related to deploying changes.
- Application issues resulting from deployments.
Adding a new label to mark these incidents as having been related to a deployment would allow us to measure the impact a lower MTTP is having on customers.
I think the change would involve:
- Create a "deployment-related' label and add to relevant incidents.
- Modify https://about.gitlab.com/handbook/engineering/infrastructure/performance-indicators/#mean-time-between-failures-mtbf to show the trend of deployment-related MTBF - later we can add in tacking incidents of other types. MTBF-deployment-related could also be a totally separate chart if you prefer.