We were testing with forcing a failure in upgrade.go to catch a condition where a test didn't run due to a previous failure

 
err != nil {
//before returning the err force a failure in a test scenario
disruption.RecordJUnit(f, "[sig-cluster-lifecycle] intentional test failure", func() (error, bool)

{ framework.Logf("Validating failure occurs with err") return fmt.Errorf("Validating failure occurs %s: %s", "upgrade", "forced failure (err)"), false }

)
recordClusterEvent(kubeClient, uid, "Upgrade", "UpgradeFailed", fmt.Sprintf("failed to acknowledge version: %v", err), true)
return err
} else {
//force the failure here as well just to make sure we get our signal
disruption.RecordJUnit(f, "[sig-cluster-lifecycle] intentional test failure", func() (error, bool)

{ framework.Logf("Validating failure occurs without err") return fmt.Errorf("Validating failure occurs %s: %s", "upgrade", "forced failure (no err)"), false }

)
}
 

The else condition above fired and the test failed. However the job succeeded. Question is how did the job succeed with a failed test.
 

https://prow.ci.openshift.org/view/gs/origin-ci-test/pr-logs/pull/27059/pull-ci-openshift-origin-master-e2e-gcp-upgrade/1519062970204688384