Background

The origin test suite does not test CPMS, so, it should never have a CPMS rollout occur during a run.

We should add a test that checks that, early in the suite, the control plane machines are all named <cluster-name>master<index>. If for any reason we see a control plane machine matching <cluster-name>master<random>-<index> we know that the CPMS has rolled out and the test should be aborted until we work out why the CPMS rolled out.

The hope here is that it becomes very obvious when there are issues with CPMS, even when these issues are introduced by other repositories.

Steps

• Add test to origin as per above description
• Discuss with TRT to make sure the test is running in regular and serial jobs and is early enough to be detect the issue if other failures may kill the job early
• Tests will be added into https://github.com/openshift/origin/tree/master/test/extended/machines

Stakeholders

• Cluster infra
• Installer
• TRT

Definition of Done

• Origin tests fail with an obvious CPMS related issue when the CPMS attempts a rollout

• Docs

• <Add docs requirements for this card>

• Testing

• <Explain testing that will be added>