The two that I looked at appear related to the etcd quorum guard pod. Diagnosis looks like this
1. inspect the co/machine-config . This shows that the pool didn't upgrade and lists a problem with "error when evicting pod" and lists a quorum guard pod
2. That quorum guard pod is for etcd and the two I looked at were both in ns/openshift-machine-config-operator
I haven't dug further.
from here: https://openshift-release.apps.ci.l2s4.p1.openshiftapps.com/#4.6.0-0.ci find one where upgrade column is red. Click the job, then click the failed upgrade.
After https://github.com/openshift/cluster-etcd-operator/pull/429 upgrade minor appears to be passing regularly.
Verified with checking ci job result.