We ended up having to remove the subscriber from the cluster (on the publisher's Administration>Server Manager>Server Configuration page), then on out-of-sync subscriber ran CLI command "cluster reset-database". Finally, we added it back to the cluster.