FY21-Q1 Engineering Productivity OKR: Increase efficiency and productivity by increasing the performance and stability of our CI pipeline => 77.6%

• Key Result: Decrease average merge request pipeline duration from 72.3 minutes to 45 minutes. => 38%, average for April is at 62 mins - gitlab-org&2530 (closed)
• Key Result: Increase master pipeline success rate from 77% to 90%. => 95%, 89% as of 2020-04-27, 10D moving average at 88% - gitlab-org&2529 (closed)
• Key Result: Create 1 performance indicators in periscope for release post rate => 100% - https://gitlab.com/gitlab-org/quality/team-tasks/-/issues/343

---

Retrospection

Good

• We made good progress on master stability and pipeline duration.
• The master broken triage process has been effective.
• We implemented and learned a lot from the Release Post rate calculation

Bad

• Priority conflict - priorities shifted in the middle of March to focus on Review App stabilization and GitLab CI Cost Reduction efforts. The later is aligned with Decrease average merge request pipeline duration from 72.3 minutes to 45 minutes. but we have prioritized efforts to reduce the number of pipelines or number of jobs executing in pipeline over saving time.
• Pipeline success rate decreased due to slow quarantine and failures during non-monitored periods (weekends).
• Using epics for work in OKR was inefficient to keep track of work compared to linked issues.
• We don't have EP coverage for Friday esp in pacific timezone. That may lead to a weekend of broken builds if no engineers step in and help.

Try

• Set expectations for engineers on Fridays until there someone to help fill in AMER Pacific timezone.
• If we are ahead on the OKRs, try to work on things that can help improve the productivity of the on-call and triage processes.