問題文
A platform team is designing the shared pipeline and is deciding how to split the work into stages. Which split makes a failure most useful to the developer who caused it?
選択肢
- Split so that each stage answers one question, and order them so that the cheapest and most frequently failing checks run first.
- Split so that every stage runs in parallel with all the others, because the total wall-clock time of the pipeline is what developers notice and reducing it is the surest way to keep them integrating frequently.
- Split by the team that owns each check, so that failures are routed to the right owner and the stage boundaries follow the organization chart of the engineering group.
- Put every check into a single stage so that the pipeline has a single point of failure to investigate and one result to report to the developer.