Cloned because the bugzilla bot doesn't like the original BZ for some reason, so trying with this new one

+++ This bug was initially created as a clone of Bug #2091730 +++

Around May 26 the MCO e2e-gcp-op tests started failing with 

=== RUN   TestMCCPausedKubeletCAAlert
    mcc_test.go:30: Pausing pool
    mcc_test.go:35: Paused
    mcc_test.go:38: Patching certificate
    mcc_test.go:41: Patched
    mcc_test.go:44: Waiting for rendered config to get stuck behind pause
    mcc_test.go:47: Certificate stuck behind paused (as expected)
    mcc_test.go:50: Getting monitoring token
    mcc_test.go:52: 
        	Error Trace:	mcc_test.go:52
        	Error:      	Expected nil, but got: &errors.errorString{s:"No token found in openshift-monitoring secrets"}
        	Test:       	TestMCCPausedKubeletCAAlert
--- FAIL: TestMCCPausedKubeletCAAlert (20.25s)
FAIL
FAIL	github.com/openshift/machine-config-operator/test/e2e	1258.101s
FAIL

In a pretty consistent fashion, see

https://prow.ci.openshift.org/job-history/gs/origin-ci-test/pr-logs/directory/pull-ci-openshift-machine-config-operator-master-e2e-gcp-op

example: https://prow.ci.openshift.org/view/gs/origin-ci-test/pr-logs/pull/openshift_machine-config-operator/3166/pull-ci-openshift-machine-config-operator-master-e2e-gcp-op/1531149248697995264