Description of problem:

It was renamed between ec.1 and ec.2:

$ oc adm release extract --to ec.1 quay.io/openshift-release-dev/ocp-release:4.15.0-ec.1-x86_64
$ oc adm release extract --to ec.2 quay.io/openshift-release-dev/ocp-release:4.15.0-ec.2-x86_64
$ yaml2json <ec.1/0000_30_cluster-api_10_webhooks.yaml | jq -r  .metadata.name
validating-webhook-configuration
$ yaml2json <ec.2/0000_30_cluster-api_10_webhooks.yaml | jq -r  .metadata.name
cluster-capi-operator

And the presence of the old config breaks updates across the gap, as the operator tries to act on resources that are still guarded by a webhook config, despite there no longer being anything serving the hooks it had pointed at. Or something like that. In any case, the cluster-api ClusterOperator goes Degraded=True on SyncingFailed with {{Failed to resync for operator: 4.15.0-ec.2 because &

}} until the old ValidatingWebhookConfiguration is deleted, and after that deletion, the ClusterOperator recovers.

Version-Release number of selected component (if applicable):

4.15.0-ec.2.

How reproducible:

Untested, but I'd guess 100%.

Steps to Reproduce:

1. Install a tech-preview 4.15.0-ec.1 cluster.
2. Request an update to 4.15.0-ec.2.
3. Wait an hour or so.

Actual results:

cluster-api ConsoleOperator is Degraded=True, blocking further progress in the ClusterVersion update.

Expected results:

ClusterVersion update happily completes.