The MCO's gcp-e2e-op-single-node job https://prow.ci.openshift.org/job-history/gs/origin-ci-test/pr-logs/directory/pull-ci-openshift-machine-config-operator-master-e2e-gcp-op-single-node has been failing consistently since early Jan.

It always fails on TestKernelArguments but that happens to be the first time where it gets the node to reboot, after which the node never comes up, so we don't get must-gather and (for some reason) don't get any console gathers either.

This is only 4.16 and only single node. Doing the same test on HA gcp clusters yield no issues. The test itself doesn't seem to matter as the next test would fail the same way if it was skipped.

This can be reproduced so far only via a 4.16 clusterbot cluster.

4.16

100%

    1. install SNO 4.16 cluster
    2. run MCO's TestKernelArguments
    3.
    

Node never comes back up

Test passes