It is now possible to configure certain jobs to execute first, and, if they pass, a second set of jobs run. This useful to save resources and only execute expensive jobs if the cheap ones pass.

We could run all integration tests at the second stage, as they are expensive - they require a dedicated OCP cluster to run. All the others (unit, lint, gen-check, etc) would run at the first stage.

Doc: https://docs.ci.openshift.org/docs/how-tos/creating-a-pipeline/