Trigger QA pipeline instead of running the QA tests in omnibus-gitlab pipeline
Right now, we run the QA tests in the omnibus-gitlab pipeline itself. This is a repetition of what QA project does in their pipeline. Also, whenever new QA scenarios are added, we will have to add them to our script also. This violates the SSOT principle we should be following.
@grzesiek has updated the trigger-build-omnibus
script in CE/EE repos so that it waits for Omnibus pipeline to complete and grabs the status of omnibus pipeline and decides whether CE/EE pipeline should pass/fail.
Proposal
- Instead of running specs in omnibus-gitlab pipeline, trigger QA pipeline as the last stage of omnibus-gitlab pipeline
- Adapt the trigger-build-omnibus script from CE/EE repo and wait for the QA pipeline to finish.
- Decide whether omnibus-gitlab pipeline passes/fails depending on the status of QA pipeline. This will in turn decide whether CE/EE pipeline passes/fails.
- Once we do this for triggers from CE/EE repositories, extend it to triggers from omnibus-gitlab repository also.