Test overview logs: https://prow.ci.openshift.org/view/gs/test-platform-results/pr-logs/pull/openshift_machine-config-operator/4448/pull-ci-openshift-machine-config-operator-master-e2e-gcp-op-techpreview/1807908151807184896

Build logs: https://gcsweb-ci.apps.ci.l2s4.p1.openshiftapps.com/gcs/test-platform-results/pr-logs/pull/openshift_machine-config-operator/4448/pull-ci-openshift-machine-config-operator-master-e2e-gcp-op-techpreview/1807908151807184896/artifacts/e2e-gcp-op-techpreview/test/artifacts/TestYumReposBuilds-build-rendered-layered-3860b34de3c240b560d6a620195007b8-image-build.log

I noticed these failures on occasion in unrelated PRs. I chatted with zzlotnik@redhat.com on slack about this, and the main theory at the moment is that image registry secret rotation takes place after the build pod had started. Although we have the controllerconfig path implemented for image registry secrets, there still seems to be some kind of timing issue.