The owner cache does not support clustering

It needs to send serializable events for the various changes which can be applied to the replication document index so that the cache can be invalidated on other cluster members.