I'm trying to create a 2 node SQL AlwaysOn group in Azure. My idea was to use a Large VM for primary and a Small VM as a secondary. Since the secondary won't really be used much except during security updates and host patches, etc, I'd rather not pay for a full Large VM to basically be a hot spare. The performance hit while the Small VM is primary is acceptable.
So far this works, but I can't seem to get any sort of failback to work. I've tried to put the Large VM as the top preferred for the availability group in the cluster but that doesn't help. I even tried making the Large VM the only node preferred, but as soon as I take the Large VM down and bring it back up, the small VM becomes the top preferred for the availability group.
I checked out the SQL 2012 forums and failback is not even recommended for SQL 2012 - they recommend manual failovers. However, since I don't have control over when the primary may be taken down, I'd really like failback to be automatic once Azure is done doing its thing...