https://testgrid.k8s.io/redhat-openshift-ocp-release-4.14-informing#release-openshift-ocp-installer-e2e-aws-csi-4.14 is an example of a job that is failing to install 100% of time.

Joel (machine-api) inspected two of the failures and found
> Workers are in provisioned, MCS has records of the workers checking in for their ignition, and then no CSRs. So something between fetching ignition and starting kubelet went wrong

Ryan was called into a couple threads, but requires aws console logs.  We don't have a way to collect those on this particular job, BUT we do have an offer to provide AWS console access.

Slack thread here