We have now restarted the Cache node on the secondary cluster. DNS records are now being returned as expected. We're still monitoring this though.

Please note: Due to DNS caching on resolvers these can take 4 hours to propagate.